Imagebased Computational Modeling Of The Human Circulatory And Pulmonary Systems Methods And Applications 1st Edition Daniel R Thedens Auth

Imagebased Computational Modeling Of The Human
Circulatory And Pulmonary Systems Methods And
Applications 1st Edition Daniel R Thedens Auth
download
https://ebookbell.com/product/imagebased-computational-modeling-
of-the-human-circulatory-and-pulmonary-systems-methods-and-
applications-1st-edition-daniel-r-thedens-auth-2010460
Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.
Imagebased Case Studies In Ent And Head Neck Surgery 1st Edition
Rahmat Omar
https://ebookbell.com/product/imagebased-case-studies-in-ent-and-head-
neck-surgery-1st-edition-rahmat-omar-48701760
Imagebased Modeling 1st Edition Long Quan
https://ebookbell.com/product/imagebased-modeling-1st-edition-long-
quan-49021764
Image Based Computing For Food And Health Analytics Requirements
Challenges Solutions And Practices Ibcfha Rajeev Tiwari
https://ebookbell.com/product/image-based-computing-for-food-and-
health-analytics-requirements-challenges-solutions-and-practices-
ibcfha-rajeev-tiwari-49436062
Imagebased Visualization 1st Edition Christophe Hurter
https://ebookbell.com/product/imagebased-visualization-1st-edition-
christophe-hurter-50783402

Imagebased Sexual Abuse A Study On The Causes And Consequences Of
Nonconsensual Nude Or Sexual Imagery Nicola Henry
https://ebookbell.com/product/imagebased-sexual-abuse-a-study-on-the-
causes-and-consequences-of-nonconsensual-nude-or-sexual-imagery-
nicola-henry-33720166
Imagebased Geometric Modeling And Mesh Generation Jessica Zhang Ed
https://ebookbell.com/product/imagebased-geometric-modeling-and-mesh-
generation-jessica-zhang-ed-4158908
Imagebased Rendering 1st Edition Heungyeung Shum Shingchow Chan
https://ebookbell.com/product/imagebased-rendering-1st-edition-
heungyeung-shum-shingchow-chan-4192282
Imagebased Fractal Description Of Microstructures 1st Edition J M Li
https://ebookbell.com/product/imagebased-fractal-description-of-
microstructures-1st-edition-j-m-li-4200590
Imagebased Multilevel Biomechanical Modeling For Fallinduced Hip
Fracture 1st Edition Yunhua Luo Auth
https://ebookbell.com/product/imagebased-multilevel-biomechanical-
modeling-for-fallinduced-hip-fracture-1st-edition-yunhua-luo-
auth-5737298

Image-Based Computational Modeling
of the Human Circulatory and Pulmonary Systems

Krishnan B. Chandran · H.S. Udaykumar ·
Joseph M. Reinhardt
Editors
Image-Based Computational
Modeling of the Human
Circulatory and Pulmonary
Systems
Methods and Applications
Foreword by Peter Hunter
1 3

Editors
Krishnan B. Chandran
Department of Biomedical Engineering
College of Engineering
1138 Seamans Center
The University of Iowa
Iowa City, IA 52242, USA
chandran@engineering.uiowa.edu
H.S. Udaykumar
Department of Mechanical
and Industrial Engineering
2408 Seamans Center
hs-kumar@uiowa.edu
Joseph M. Reinhardt
Department of Biomedical Engineering
1402A Seamans Center
jmr@engineering.uiowa.edu
ISBN 978-1-4419-7349-8 e-ISBN 978-1-4419-7350-4
DOI 10.1007/978-1-4419-7350-4
Springer New York Dordrecht Heidelberg London
© Springer Science+Business Media, LLC 2011
All rights reserved. This work may not be translated or copied in whole or in part without the written
permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York,
NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in
connection with any form of information storage and retrieval, electronic adaptation, computer software,
or by similar or dissimilar methodology now known or hereafter developed is forbidden.
The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are
not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject
to proprietary rights.
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

To
Vanaja, Aruna and Kelly, Anjana and Jaime
KBC
To
H. N. S. Murthy
HSU
To
Jennifer, Eliza, and William
JMR

Foreword
This book is very timely. The medical imaging community has long used
numerical techniques for extracting anatomical structure from clinical images, but
the use of anatomically and biophysically based computational models to interpret
physiological function in a clinical setting is relatively new. The book is focused
on the cardiovascular and pulmonary systems but the imaging and computational
approaches discussed here are equally applicable across many other organ systems.
Many of the authors are experts in clinical image analysis, as well as com-
putational methods, so not surprisingly the starting point for modeling clinical
structure–function relations is often an image of some sort—MRI, CT, or ultra-
sound for anatomical imaging and PET or SPECT for functional images. Part I
therefore reviews image acquisition and analysis for all of these techniques. Part II
then deals with the physics and computation of soft tissue mechanics, fluid mechan-
ics, and fluid–structure interaction. Multi-scale approaches for understanding blood
flow mechanics are also discussed in this section. Part III focuses on the use of
image-based computational analysis of cardiopulmonary disease. Applications to
diagnostics, therapeutics, surgical planning, and the design of medical devices are
considered throughout the book.
The computational methods described and used here include the continuum-
based finite element and finite difference techniques, familiar to engineers, and the
particle-based lattice Boltzmann methods more familiar to applied mathematicians.
The corresponding equations (e.g., Navier–Stokes, Fokker–Planck) are all derived
from the same underlying physical laws: the advantages of the former are partly
to do with the widespread availability of appropriate constitutive laws and highly
developed computer codes; the advantages of the latter are more apparent when
dealing with multi-scale physics in which material continuum properties emerge
out of the statistical behavior of interacting particles.
Material properties are a recurring theme in this book and, in fact, in any book
dealing with an engineering physics analysis of function, because at the level of
tissues the properties of biological materials such as soft tissue or blood (and
their characterization via constitutive laws) are essential summaries of complex
structure–function relations at smaller spatial scales. Several chapters address the
vii

viii Foreword
important question of how to derive material properties with meso-scale mod-
eling from our knowledge of the structure and function of tissue components.
One reason why this is such an important task is because diseases are processes
operating mechanistically at the protein level but manifest clinically at the tissue
level. Understanding this multi-scale connection and, in particular, the growth and
remodeling of tissue under changed loading environments is therefore vital.
This book is an immensely valuable contribution to the computational analy-
sis of structure–function relations in health and disease. It is relevant to current
clinical medicine but, perhaps more importantly, provides a guide to computa-
tional approaches that will undoubtedly underpin future evidence-based treatment
of disease.
Auckland Bioengineering Institute Peter Hunter FRS
Auckland, New Zealand

Preface
Physiological processes in living systems involve the complex interactions of
electrical activity, chemical reactions, and physical phenomena such as mass,
momentum, and energy transport. Deviations of any of these processes from their
normal states may result in the initiation of diseases. A thorough understanding of
physiological processes as they occur in the normal, healthy state as well as under
pathological conditions is necessary so that diseases can be detected early enough
for interventions to be efficacious. An understanding of complex physiological func-
tions is also vital in the design and development of implants such as vascular stents
and heart valve prostheses in human circulation and similar devices in other organ
systems. In vivo and in vitro experimental studies, and in recent decades computer
simulations, have provided valuable insight into the complex functional physiology
and pathophysiology, and our knowledge in these areas continues to grow rapidly.
However, in vivo experiments in human subjects and animals require ethical consid-
erations, and the data obtained from in vivo experiments are limited due to practical
considerations. In vitro experiments often require expensive equipment such as par-
ticle image velocimetry (PIV) systems, and yet there remain limitations in data
acquisition due to restrictions of optical access to the areas of interest.
Recent advances in medical imaging instruments, such as magnetic resonance
(MR), computed tomography (CT), and ultrasound imaging systems, have improved
both the spatial and temporal resolution of the image data that can be acquired. With
the appropriate acquisition protocol, these instruments can acquire 3D (volumetric)
and even 4D (volume data plus time) data with exquisite anatomic detail. The image
data can be visualized using computer graphics techniques to show geometric infor-
mation and can be processed to provide realistic anatomic models for subsequent
computer simulations that explore physiologic function.
With the advent of high-speed computers, computational simulations are increas-
ingly playing a major role in our ability to analyze the physiological processes in
the visceral organs and in the human musculoskeletal system. Computational simu-
lations, with appropriate experimental validation, are being increasingly employed
for various applications in human health care and have enabled us to reduce the num-
ber of animal models required for such studies. It is clear, however, that modeling
of biological systems is an extremely challenging enterprise, given the complexity
ix

x Preface
of such systems and the essential roles played by genetic factors and biological
variability. Therefore, while a truly “accurate” model of a physiological system or
process is very difficult to achieve, there is immense value in developing computa-
tional models that can capture essential features of the behavior of a system under
well-defined physicochemical conditions. Computer simulations (1) are relatively
inexpensive; (2) can cover wide ranges of parameter spaces; (3) can be improved
over time with improved inputs and other information from experiments or with
advances in modeling techniques, numerical methods, and computer hardware; and
(4) can provide information on flow and stress fields that are difficult to measure or
visualize.
The development of computational techniques and advances in hardware in terms
of speed and memory have therefore established computer simulations as a strong
source of knowledge regarding the behavior of biological systems. In fact, the cur-
rent phase of computational developments is directed toward enabling increasingly
sophisticated representations of biological systems. A particular case is that of
multi-scale modeling of such systems. Physiological processes in living systems
vary over a wide range of temporal and spatial scales. For example, chemical reac-
tions that take place at a subcellular level require analysis at a timescale on the order
of nanoseconds and at spatial dimensions on the order of nanometers. On the other
hand, functional physiology of visceral organs such as the human heart involves a
timescale on the order of seconds and at dimensions on the order of centimeters.
Disease processes such as atherosclerosis, a common arterial disease in humans,
develop during a time span of several years. Computational simulations on spatial
and temporal scales ranging from nanometers to meters and nanoseconds to years
are continuing to be developed, and strategies for integrating both spatial and tempo-
ral scales are being explored. In the last five decades, the explosion of new imaging
modalities for structural and functional imaging of organs in the human body has
also provided additional information for simulations attempting to model complex
anatomy and physiology. It can be anticipated that computational simulations will
increasingly play a vital role in the area of human health care.
In this book, we address the current status and possible future directions of
simulations that have been employed and are continuing to be developed for appli-
cations in the human cardiovascular and pulmonary systems. In these two systems,
simulations involve the description of the complex fluid flow (blood flow in the car-
diovascular system and air flow in the pulmonary system), the mechanics of the soft
tissue (vessel and airway walls, cardiac structures, and lung tissue), and the con-
stant interaction between fluids and soft tissue. Typical disease processes, such as
atherosclerosis in the human arteries and emphysema in the human lungs, result
from alterations at the microstructural level with alterations in viscous properties
and mass transport within local regions. Realistic simulation of the physiology and
alterations resulting in the initiation and development of disease processes requires
the following:
a. Acquisition of images of the organs of interest employing appropriate imaging
modality, employment of state-of-the-art image processing and segmentation,

Preface xi
and reconstruction of morphologically realistic three-dimensional (3D) geome-
try of the region of interest as a function of time.
b. Appropriate boundary conditions (pressures, flow rates, etc.) obtained from
physiological measurements.
c. Development of computational techniques for the fluid flow (e.g., to represent
blood rheology in the human circulation and turbulent compressible flow analy-
sis for transport of air in the lung airways), the soft tissue (nonlinear anisotropic
material description for the cardiac and blood vessel structures and the pul-
monary airways from the trachea to the alveolar sacs), and the fluid–structural
interaction analyses.
d. Validation of the computational techniques with appropriate experimental or
computational simulations, before the application of the simulations, to describe
the various physiological and pathophysiological processes.
The chapters to follow in this work are divided into three sections:
Part I deals with image data acquisition and geometric reconstruction commonly
employed in the diagnosis and treatment of cardiovascular and pulmonary diseases.
Chapter 1 discusses commonly employed imaging modalities used for anatomical
and functional imaging of these two-organ systems, as well as trade-offs between
spatial and temporal resolution, invasiveness of the imaging technique, and the use
of ionizing vs. non-ionizing radiation. Chapter 2 focuses on contemporary image
analysis and data processing techniques in order to identify anatomic structures in
the images, delineate region boundaries, and construct three-dimensional geometric
representations of regions of interest to be employed in the simulations.
Part II consists of discussions of state-of-the-art computational techniques for
biological soft tissue, biological fluid, and the analysis of interaction between the
fluid and the surrounding tissue. Chapter 3 presents the numerical approaches for
solving the Navier–Stokes equations at two distinct scales, viz., the large-scale sys-
tem that applies at the level of large blood vessels and prosthetic devices, and the
small-scale systems that apply to the microvasculature. Chapter 4 details the mod-
eling and solution of the equations governing the dynamics of soft tissue in the
cardiovascular system. Chapter 5 focuses on the issue of fluid–structure interactions
and distinguishes three types of techniques used to simulate the presence of struc-
tures immersed in blood flow. Issues pertaining to the behavior of the fluid–structure
coupled solutions as they are influenced by the properties of the immersed solid are
discussed.
The majority of the simulations published to date are focused mainly at the organ
level where the biological soft tissue as well as the fluid can be treated as a contin-
uum. There are limitations imposed on such simulations due to various practical
constraints, including computer memory, processing speed, modeling uncertainties
and complexity, biological variability. Even with the increasing speeds and memory
densities of state-of-the-art computers, with the finest possible mesh density in the
computational simulations, organ systems can at best be resolved down to dimen-
sions in the order of millimeters—i.e., cellular and subcellular phenomena need to

xii Preface
be modeled. However, in the last several decades, our knowledge of the physio-
logical functions and pathological processes at the cellular and subcellular levels
has also increased significantly. On the horizon of the computational landscape lies
the possibility of linking computational analyses from the organ level (i.e., at the
length scale of meters) all the way through to the cellular and subcellular levels (at
the length scales of microns) and in time from nanoseconds to disease evolution
scales. For example, numerous studies have focused on the relationship between
the shear stress induced by the blood flow on the endothelial cells and the shear
stresses computed by the simulations at the various arterial segments, as these are
related to morphologically observed sites of atherosclerotic plaque development.
Numerous experimental studies and simulations have also been employed at the
level of endothelial cells in order to understand the response of the cells to external
stimuli in the form of structural changes as well as to understand chemical alter-
ations and the release of various growth factors and other enzymes. Recognizing
that it is beyond the capabilities of even state-of-the-art high-performance comput-
ers to incorporate events at the subcellular level to those at the organ level through
direct numerical computations, multi-scale simulation techniques are being inves-
tigated. Chapter 6 attempts to sketch the outlines of such a multi-scale modeling
effort as it applies to the transport of blood at the micro- and mesoscales. The chal-
lenge of connecting these efforts to the large-scale blood flow simulations detailed
in Chapters 3 and 5 lies at the frontier of multi-scale modeling.
The focus of Part III is on the application of computational simulations to a
range of problems often encountered in the human circulatory and pulmonary sys-
tems. Chapter 7 addresses the current status of the simulations on our understanding
of the arterial blood flow and the relationship between fluid-induced stresses and
atherosclerotic plaque development. Topics include three-dimensional reconstruc-
tion of coronary arterial segments and simulation of coronary flow dynamics, flow
simulations in the aorta and arterial bifurcations, and image-based simulation of
abdominal aortic aneurysms (AAA). Models to analyze the endovascular implants
for treating AAA and bypass grafting for the treatment of arterial occlusions are
also discussed in this chapter. Detailed treatment of the biomechanics of both AAA
and cerebral aneurysms is the topic of Chapter 8. The biomechanical modeling of
aneurysm segments includes the effect of the material property of diseased arte-
rial segments and prediction of rupture of aneurysms. The effect of alterations in
the fluid flow on the biomechanics of the aneurysms is also discussed in detail
in this chapter. Chapter 9 deals with the application of computational simulations
for interventional treatments. Topics addressed in this chapter include application
of modeling and simulation to assess atheromatous plaque vulnerability to rupture,
mechanical effects of balloon angioplasty, and the design of endovascular stents that
are implanted after angioplasty to open occluded arterial segments. In the second
part of the chapter, the use of simulations for the surgical planning of single ven-
tricle heart defect (SHVD) is described. As discussed with specific examples, rapid
development of other patient-specific applications on interventional techniques and
surgical planning is anticipated in the near future. The focus of Chapter 10 is on

Preface xiii
the application of modeling and simulation toward an understanding of the biome-
chanics of the respiratory system and the complex relationships between pulmonary
anatomy, tissue dynamics, and respiratory function, as well as on how these rela-
tionships can change in the presence of pathological processes. Chapters 11 and
12 deal with the biomechanics of the heart valve function. The native heart valves
have a complicated three-dimensional geometry. Since diseases of the valves are
predominant in the left heart, the functional biomechanics of the aortic and mitral
valves are of interest in increasing our understanding of the function of the healthy
valve, the mechanical factors that contribute to the valvular diseases—such as calci-
fication of the leaflets—and valvular regurgitation. These dynamic simulations have
potential applications in the planning of patient-specific valvular repair strategies as
well as in the development of tissue-engineered valve replacements. The dynamic
simulations of the heart valves are challenging, requiring the inclusion of the entire
valvular apparatus including the annulus, leaflets, and the ascending aorta for the
aortic valves, and the leaflets, annulus, chordae and the papillary muscles for the
mitral valves, as well as the development of accurate fluid–structure interaction anal-
ysis. These topics are covered in Chapter 11 along with the potential applications
on the improved designs for biological valve prostheses. Simulations to understand
the cause of thrombus deposition, a continuing and significant problem associated
with mechanical valve prostheses, and the simulations toward our understanding of
the fluid mechanical factors responsible for the same is the topic of Chapter 12.
Iowa City, Iowa Krishnan B. Chandran
H.S. Udaykumar
Joseph M. Reinhardt

Contents
Part I Cardiac and Pulmonary Imaging, Image Processing,
and Three-Dimensional Reconstruction in
Cardiovascular and Pulmonary Systems
1 Image Acquisition for Cardiovascular and Pulmonary
Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Daniel R. Thedens
2 Three-dimensional and Four-dimensional
Cardiopulmonary Image Analysis . . . . . . . . . . . . . . . . . . 35
Andreas Wahle, Honghai Zhang, Fei Zhao, Kyungmoo Lee,
Richard W. Downe, Mark E. Olszewski, Soumik Ukil,
Juerg Tschirren, Hidenori Shikata, and Milan Sonka
Part II Computational Techniques for Fluid and Soft
Tissue Mechanics, Fluid–Structure Interaction,
and Development of Multi-scale Simulations
3 Computational Techniques for Biological Fluids: From
Blood Vessel Scale to Blood Cells . . . . . . . . . . . . . . . . . . . 105
Fotis Sotiropoulos, Cyrus Aidun, Iman Borazjani,
and Robert MacMeccan
4 Formulation and Computational Implementation
of Constitutive Models for Cardiovascular Soft Tissue
Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Michael S. Sacks and Jia Lu
5 Algorithms for Fluid–Structure Interaction . . . . . . . . . . . . . 191
Sarah C. Vigmostad and H.S. Udaykumar
xv

xvi Contents
6 Mesoscale Analysis of Blood Flow . . . . . . . . . . . . . . . . . . . 235
Jeffrey S. Marshall, Jennifer K.W. Chesnutt,
and H.S. Udaykumar
Part III Applications of Computational Simulations
in the Cardiovascular and Pulmonary Systems
7 Arterial Circulation and Disease Processes . . . . . . . . . . . . . . 269
Tim McGloughlin and Michael T. Walsh
8 Biomechanical Modeling of Aneurysms . . . . . . . . . . . . . . . . 313
Madhavan L. Raghavan and David A. Vorp
9 Advances in Computational Simulations for Interventional
Treatments and Surgical Planning . . . . . . . . . . . . . . . . . . 343
Diane A. de Zélicourt, Brooke N. Steele, and Ajit P. Yoganathan
10 Computational Analyses of Airway Flow and Lung Tissue
Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
David W. Kaczka, Ashley A. Colletti, Merryn H. Tawhai,
and Brett A. Simon
11 Native Human and Bioprosthetic Heart Valve Dynamics . . . . . . 403
Hyunggun Kim, Jia Lu, and K.B. Chandran
12 Mechanical Valve Fluid Dynamics and Thrombus Initiation . . . . 437
Tom Claessens, Joris Degroote, Jan Vierendeels,
Peter Van Ransbeeck, Patrick Segers, and Pascal Verdonck
Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463

Contributors
Cyrus Aidun George W. Woodruff School of Mechanical Engineering, Georgia
Institute of Technology, Atlanta, GA 30332, USA, cyrus.aidun@me.gatech.edu
Iman Borazjani Mechanical and Aerospace Engineering Department, SUNY
University at Buffalo, Buffalo, NY 14260, USA, iman@buffalo.edu
Krishnan B. Chandran Department of Biomedical Engineering, College of
Engineering, 1138 Seamans Center, The University of Iowa, Iowa City, IA 52242,
USA, chandran@engineering.uiowa.edu
Jennifer K.W. Chesnutt Department of Mechanical Engineering, The University
of Texas at San Antonio, San Antonio, TX, USA, jennifer.chesnutt@utsa.edu
Tom Claessens BioMech Research Group, Faculty of Applied Engineering,
Department of Mechanics, University College Ghent, B-9000 Ghent, Belgium,
tom.claessens@hogent.be
Ashley A. Colletti University of Toledo School of Medicine, Toledo, OH 43614,
USA, ashley.colletti@rockets.utoledo.edu
Joris Degroote Faculty of Engineering, Department of Flow, Heat and Combustion
Mechanics, Ghent University, B-9000 Ghent, Belgium, joris.degroote@ugent.be
Richard W. Downe Department of Electrical and Computer Engineering, Iowa
Institute for Biomedical Imaging, The University of Iowa, Iowa City, IA 52242,
USA, richard-downe@uiowa.edu
David W. Kaczka Harvard Medical School, Beth Israel Deaconess Medical
Center, Boston, MA 02215, USA, dkaczka@bidmc.harvard.edu
Hyunggun Kim Division of Cardiology, Department of Internal Medicine, The
University of Texas Health Science Center at Houston, Houston, TX 77030, USA,
hyunggun.kim@uth.tmc.edu
Kyungmoo Lee Department of Electrical and Computer Engineering, Department
of Biomedical Engineering, Iowa Institute for Biomedical Imaging, The University
of Iowa, Iowa City, IA, kyungmoo-lee@uiowa.edu
xvii

xviii Contributors
Jia Lu Department of Mechanical and Industrial Engineering, The University of
Iowa, 2137 Seamans Center, Iowa City, IA 52242, jia-lu@uiowa.edu
Robert MacMeccan George W. Woodruff School of Mechanical Engineering,
Georgia Institute of Technology, Atlanta, GA 30332, USA,
robert.macmeccan@milliken.com
Jeffrey S. Marshall School of Engineering, University of Vermont, Burlington,
VT 05405, USA, jeffm@cems.uvm.edu
Tim McGloughlin Department of Mechanical and Aeronautical Engineering,
Materials and Surface Science Institute, Centre for Applied Biomedical
Engineering Research, University of Limerick, National Technological Park,
Castletroy, Limerick, Ireland, tim.mcgloughlin@ul.ie
Mark E. Olszewski Philips Healthcare, CT Clinical Science, Highland Heights,
OH, USA, mark.olszewski@philips.com
Madhavan L. Raghavan Department of Biomedical Engineering, 1136 Seamans
Center, College of Engineering, The University of Iowa, Iowa City, IA
52242-1527, USA, ml-raghavan@uiowa.edu
Michael S. Sacks Department of Bioengineering, Swanson School of Engineering,
University of Pittsburgh, Pittsburgh, PA 15219, USA; School of Medicine, The
McGowan Institute, University of Pittsburgh, Pittsburgh, PA 15219, USA,
msacks@pitt.edu
Patrick Segers bioMMeda Research Group, Faculty of Engineering, Ghent
University, B-9000 Ghent, Belgium, patrick.segers@ugent.be
Hidenori Shikata Ziosoft Inc., Redwood City, CA, USA, hidenori@shikatas.net
Brett A. Simon Harvard Medical School, Beth Israel Deaconess Medical Center,
Boston, MA 02215, USA, bsimon@bidmc.harvard.edu
Milan Sonka Department of Electrical and Computer Engineering, Iowa Institute
for Biomedical Engineering, The University of Iowa, Iowa City, IA 52242, USA
Fotis Sotiropoulos St. Anthony Falls Laboratory, Department of Civil
Engineering, University of Minnesota, Minneapolis, MN 55414, USA,
fotis@umn.edu
Brooke N. Steele 2148 Burlington Nuclear Engineering Laboratories, NC State
University, Raleigh, NC 27695-7115, USA, bnsteel@gmail.com
Merryn H. Tawhai Auckland Bioengineering Institute, The University of
Auckland, Auckland, New Zealand, m.tawhai@auckland.ac.nz

Contributors xix
Daniel R. Thedens Department of Radiology, 0446 John W. Colloton Pavilion,
The University of Iowa, Iowa City, IA 52242, USA, dan-thedens@uiowa.edu
Juerg Tschirren Department of Electrical and Computer Engineering, Iowa
USA, juerg@vidadiagnostics.com
H.S. Udaykumar Department of Mechanical and Industrial Engineering, 2408
Seamans Center, College of Engineering, The University of Iowa, Iowa City, IA
52242, USA, hs-kumar@uiowa.edu
Soumik Ukil Imaging and Video Services, Nokia India Pvt. Limited, Bangalore,
India, soumik.ukil@nokia.com
Peter Van Ransbeeck BioMech Research Group, Faculty of Applied Engineering,
Department of Mechanics, University College Ghent, Ghent, B-9000, Belgium,
peter.vanransbeeck@ugent.be
Pascal Verdonck bioMMeda Research Group, Faculty of Engineering, Ghent
University, B-9000, Ghent, Belgium, pascal.verdonck@ugent.be
Jan Vierendeels Faculty of Engineering, Department of Flow, Heat and
Combustion Mechanics, Ghent University, Ghent, B-9000, Belgium,
jan.vierendeels@ugent.be
S.C. Vigmostad Department of Biomedical Engineering, 1420 Seamans Center,
College of Engineering, The University of Iowa, Iowa City, IA 52242, USA,
svigmost@engineering.uiowa.edu
David A. Vorp Departments of Surgery and Bioengineering, University of
Pittsburgh, Pittsburgh, PA 15219, USA, vorpda@upmc.edu
Andreas Wahle Department of Electrical and Computer Engineering, Iowa
USA, andreas-wahle@uiowa.edu
Michael T. Walsh Department of Mechanical and Aeronautical Engineering,
Materials and Surface Science Institute, Centre for Applied Biomedical
Engineering Research, University of Limerick, National Technological Park,
Castletroy, Limerick, Ireland, michael.walsh@ul.ie
Ajith P. Yoganathan Wallace H. Coulter School of Biomedical Engineering,
Georgia Institute of Technology and Emory University, 313 Ferst Drive, Atlanta,
GA 30332-0535, USA, ajit.yoganathan@bme.gatech.edu

xx Contributors
Diane A. de Zélicourt Petit Institute of Bioengineering, Georgia Institute of
Technology and Emory University, Atlanta, GA 30332, USA,
diane.dezelicourt@gmail.com
Honghai Zhang Department of Electrical and Computer Engineering, Iowa
USA, honghai-zhang@uiowa.edu
Fei Zhao Department of Electrical and Computer Engineering, Iowa Institute for
Biomedical Imaging, The University of Iowa, Iowa City, IA 52242, USA,
zhaof@ge.com

Part I
Cardiac and Pulmonary Imaging, Image
Processing, and Three-Dimensional
Reconstruction in Cardiovascular and
Pulmonary Systems

Chapter 1
Image Acquisition for Cardiovascular
and Pulmonary Applications
Daniel R. Thedens
Abstract Medical imaging hardware can provide detailed images of the cardiac
and pulmonary anatomy. High-speed imaging can be used to acquire time sequences
showing tissue dynamics or can capture a snapshot of the changing anatomy at an
instant in time. Some imaging modalities can also provide functional information,
such as perfusion, ventilation, and metabolic activity, or mechanical information,
such as tissue deformation and strain. With the appropriate acquisition protocol,
some of these imaging devices can acquire 3D (volumetric) and even 4D (volume
data plus time) data with excellent anatomic detail. This image data can be visu-
alized using computer graphics techniques to show geometric information, and the
data can be processed to provide realistic anatomic models for subsequent com-
puter simulations that explore physiologic function. This chapter describes the most
commonly used imaging modalities for cardiovascular and pulmonary applications
and describes some of the advantages and disadvantages of the different modalities.
New, emerging modalities that may be important imaging tools in the future are
introduced.
1.1 Introduction to Imaging
The field of medical imaging has its origin in the discovery of x-rays by Wilhelm
Roentgen in 1895, a feat which earned him the first Nobel Prize in Physics in 1901.
The very first x-ray image was of the hand of his wife. By the first decade of the
twentieth century, x-rays were being used for medical diagnosis, and the specialty of
radiology was established. X-ray imaging remained essentially the only diagnostic
imaging technique available until the 1950s.
D.R. Thedens (B)
Department of Radiology, 0446 John W. Colloton Pavilion, The University of Iowa, Iowa City,
IA 52242, USA
e-mail: dan-thedens@uiowa.edu
3
K.B. Chandran et al. (eds.), Image-Based Computational Modeling of the Human
Circulatory and Pulmonary Systems, DOI 10.1007/978-1-4419-7350-4_1,
C
Springer Science+Business Media, LLC 2011

4 D.R. Thedens
The postwar produced a proliferation of new imaging techniques as assistive
technologies in electronics and computers were applied. Nuclear medicine, which
introduces radioactive elements into the body and uses a gamma camera to detect
their distribution in the organs, appeared in the 1950s. The principles of sonar, an
important technology developed during World War II, were applied to diagnostic
imaging in the form of ultrasound imaging in the 1960s and have been a staple of
diagnostic imaging ever since.
Advances in computer technology throughout the 1970s and 1980s brought an
explosion of tomographic imaging techniques. Computed tomography (CT) used
x-ray imaging as a basis to generate two-dimensional (2D) tomographic “slices”
of the body, which eliminated many of the limitations of x-ray projection imag-
ing. Magnetic resonance imaging (MRI) similarly took the underlying technique of
nuclear magnetic resonance (NMR) to generate true 2D and three-dimensional (3D)
images of the body. MRI is distinguished by the wide variety of contrast mech-
anisms that can be generated in an exam with a single scanner. The inventors of
each of these modalities were honored with a Nobel Prize (Godfrey Hounsfield and
Allan Cormack for CT in 1979, and Paul Lauterbur and Peter Mansfield for MRI
in 2003).
All of these imaging modalities have continued to benefit from advances in
computer technology over the past 25 years. The speed and quality of the images
produced by ultrasound, CT, and MRI have improved by orders of magnitude since
their initial development, and this trend is likely to continue into the foreseeable
future. As a result, the value and importance of diagnostic imaging will continue to
grow and expand into new areas of application.
Among the fundamental differences between the various imaging modalities
considered in this chapter is the type of energy used. Imaging methods based on
x-rays, such as CT, utilize electromagnetic waves of very high energy. The energy
level is sufficiently high to cause changes and damage to living tissues. X-rays are
thus an example of ionizing radiation, and there is a risk of long-term damage to
organs if the exposure to this form of radiation is too great. MRI also uses electro-
magnetic energy in the formation of images, but the energy level used is orders
of magnitude lower, corresponding to the radiofrequency range of the spectrum
(roughly the same range as the FM radio band). As a result, the energy in MRI
is non-ionizing and cannot inflict damage on tissues. Ultrasound uses mechanical
energy in the form of high-frequency sound waves, which is also considered non-
ionizing as no tissue damage is done. Both MRI and ultrasound are safe for repeated
use, whereas exposure to x-rays should be limited due to the potentially harmful
effects.
A wide variety of diagnostic information can be generated from all of these imag-
ing modalities. Broadly, the types of information acquired can be categorized as
anatomic imaging or functional imaging. Anatomic imaging is concerned with the
depiction and distinguishing of the anatomical structures in the body. The informa-
tion may be qualitative in terms of the appearance of normal or abnormal tissues
or quantitative by measuring the size, shape, and density of the body tissues and

1 Image Acquisition for Cardiovascular and Pulmonary Applications 5
organs. Functional imaging is concerned with measuring physiologic activity such
as metabolism, blood flow, or chemical changes. Some functional measures may
be derived from anatomic data, such as calculation of cardiac function metrics like
ejection fraction from a time series of anatomic images.
1.1.1 Invasive Techniques
While noninvasive imaging techniques such as CT and MRI progress to produce
ever more detailed views of anatomy and function, the “gold standard” imag-
ing techniques for some clinical questions remain more invasive methods. For
cardiovascular assessment, x-ray angiography is still the preferred technique for
diagnosing disease of the vasculature.
The procedure involves introduction of a catheter into a major vessel, which is
then guided toward the vessels of interest. A radiodense contrast dye is injected
into the bloodstream and continuously imaged with x-ray imaging. The flow of the
contrast agent can be followed to locate areas of vessel blockage or narrowing. The
direct targeting of the vessels of interest with the contrast injection and x-ray imag-
ing means that the vessel can be seen with exquisite detail and the flow (or lack of it)
is viewable and reviewable in real time. Frequently the diagnosis and treatment of
such blockages can be performed at the same time. However, it must be recognized
that a significant number of these procedures will result in a negative diagnosis,
exposing the patient to unnecessary risks from the invasive nature of the procedure.
1.1.2 Role of Noninvasive Imaging
Because of the risks of invasive procedures, noninvasive imaging techniques based
on ultrasound, CT, or MRI, which do not involve any sort of surgical procedure,
have been widely developed to replace more and more of these previously inva-
sive techniques. Each of these modalities operates on different physical principles.
Ultrasound uses acoustic waves to probe tissue characteristics and movement. CT
forms 2D and 3D images from x-ray projections, thus depicting density. MRI images
the distribution and characteristics of hydrogen protons throughout the body, which
is essentially a map of the water content within the body. The three modalities
represent different trade-offs in acquisition time and complexity, resolution, image
quality, and versatility. All three remain important tools for diagnosing and assess-
ing treatment of cardiopulmonary disease, in many cases providing complementary
information.
While the ultimate dream of a single “one-stop shop” protocol for cardiac and
pulmonary imaging for complete assessment of the cardiopulmonary system has not
yet materialized, all three of the primary imaging techniques continue to thrive and

6 D.R. Thedens
advance. The result has been more detailed and higher quality diagnostic informa-
tion available with a drastic reduction in the need for surgical or invasive procedures.
In the sections that follow, a description of the principles of each of these modalities
as well as their roles in cardiovascular and pulmonary assessment will be presented.
1.2 Ultrasound/Echocardiography
Ultrasound imaging has its origins in the research and development that produced
advances in underwater sonar (an acronym for SOund Navigation And Ranging)
for detection of submarines. Ultrasound itself refers to acoustic or sound waves
at frequencies that are above the range of human hearing. Like all of the imaging
modalities described here, ultrasound has advanced tremendously since its original
inception. Early diagnostic techniques provided only one-dimensional (1D) profiles
of body tissues. But with the development of more sophisticated equipment and data
processing, ultrasound has become capable of 2D real-time imaging of anatomy as
well as flow and velocity, and even of 3D imaging. The portability, safety, and low
cost of ultrasound has created an ever-widening set of applications for its use both
as an initial screening tool and for quantitative assessment of morphology, function,
and flow throughout the cardiovascular system.
1.2.1 Principles of Ultrasound
This section will describe the basic principles and techniques commonly used in
ultrasound imaging. A thorough treatment of the physics of ultrasound imaging
is given in [1, 2]. The propagation of acoustic energy in body tissues depends on
its material properties, and ultrasound imaging uses the propagation and reflec-
tion of this energy to build tomographic images of these acoustic waves and their
interactions with body tissues.
Fundamentally, an acoustic wave is created by mechanical compression and rar-
efaction (“stretching”) of an elastic medium. An acoustic wave is created by a
continuous cyclic “back and forth” motion of a transducer that begins these alternat-
ing compressions and rarefactions. This pattern of displacement will then propagate
through the medium with a characteristic velocity for that material. The wave propa-
gates by transferring the mechanical energy of the compressions to adjacent material
in the medium, so that the actual movement of material as the wave moves is very
small.
In ultrasound imaging, a transducer is used to produce the high-frequency acous-
tic waves as well as to detect the signal that returns to produce the images. The
transducer consists of a piezoelectric material that can convert electrical energy into
mechanical vibrations that create the acoustic wave, with frequencies typically in
the range of 3–15 MHz. The transducer also functions as a detector, converting
the returning acoustic waves back into electrical energy for image formation, either

directly or via digitization of the signal and subsequent processing. For imaging,
the ultrasound probe typically contains an array of up to 512 such transducers in a
variety of configurations.
The velocity c at which sound propagates through a material is determined by its
bulk modulus B (related to its stiffness and compressibility) and density ρ (mass per
unit volume) as
c =

B
ρ
(1.1)
The speed of sound in air (which is highly compressible) is about 330 m/s,
whereas the velocities in soft tissues are on the order of 1800 m/s, and in a stiff
material such as bone they rise to 4000 m/s. Another important tissue characteristic
is the acoustic impedance Z, measured in kg/m2 s and defined as
Z = ρc (1.2)
with ρ and c the material density and speed of propagation, respectively. Again, the
acoustic impedance varies widely for differing materials, from 0.4×103 kg/m2 s for
air to approximately 1.5×106 kg/m2 s for many soft tissues and 7.8×106 kg/m2 s
for bone.
The formation of ultrasound images relies on the interactions of acoustic waves
occurring at the boundaries of different tissues (Fig. 1.1). When an acoustic wave
reaches a boundary between tissues with differing acoustic impedances, some of
the energy of the wave will be reflected, and this reflected energy returns back to the
source of the acoustic wave, where it can be measured and a projection or image can
be reconstructed. For two tissues of impedance Z 1 and Z 2, the fraction of reflected
intensity at a particular interface is given as
RI =

Z2 − Z1
Z2 + Z1
2
(1.3)
RI will thus range from zero (no reflection) to one (complete reflection and no
transmission).
Fig. 1.1 An ultrasound transducer transmits a mechanical wave into the tissue. When the wave
is incident on a boundary between tissues of different density ρ1 and ρ2, part of the wave is
transmitted into the second tissue and part of the wave is reflected back toward the transducer. The
reflected energy is detected and used to form a projection or an image

8 D.R. Thedens
This reflected energy from the tissue boundaries is necessary to detect changes
in the tissues. Between different types of soft tissues, such as fat and muscle, the
amount of reflected energy is small (on the order of 1%). This is sufficient to detect
and display, while the majority of the energy in the beam continues to propagate for-
ward. Boundaries where there is a large difference in impedances (such as between
air or lung and soft tissues) will result in nearly all of the energy being reflected and
none of it propagating further. As a result, material beyond air pockets and solid
materials such as bone are unobservable with ultrasound. For observing body tis-
sues, a path to the anatomy of interest must be found that does not travel through an
air-filled space, and a conducting gel is used on the face of the ultrasound transducer
to eliminate air pockets between the transducer and the skin for just this reason.
In addition to reflection, several other types of interactions between acoustic
waves and tissues may occur. Refraction occurs when the incident acoustic wave
is not perpendicular to the boundary between tissues. The result is a change in the
direction of the beam and a violation of the assumption that the wave reflection is
in a straight line, which can cause artifacts in the resulting image. In body tissues
containing very small structural elements (on the order of the wavelength of the
acoustic wave), the reflections become diffuse and scatter. This yields a “rough”
appearance to the tissue boundaries. However, since most organs have a very char-
acteristic structure, the pattern of scatter will be distinctive for specific organs, and
this can yield diagnostically important information about the tissue. Further, since
the scatter depends on the wavelength of the acoustic wave, adjusting the frequency
of the ultrasound beam can provide additional tissue characterization based on a
characteristic “texture” pattern in the tissue appearance. Finally, not all of the energy
of the ultrasound beam will be transmitted and reflected through the tissue. Some
of the energy is lost as heat in the tissues, and the beam is said to be attenuated.
The degree of attenuation is roughly proportional to the frequency of the beam and
also varies with the type of tissue, meaning that this effect can also be of diagnostic
value.
1.2.1.1 M-Mode
The fundamental method of image formation in ultrasound imaging is the pulse echo
method. A pulse of ultrasonic energy (with duration on the order of 1 μs) is rapidly
generated and transmitted into the body tissues. As described above, the interactions
of the acoustic wave with tissues of differing acoustic properties produce reflections
and scatter; these are then returned to and detected by the transducer after the pulse
transmission has been turned off. This process of transmission and signal recording
is repeated anywhere from 500 to 15,000 times per second. Each detected signal is
then processed and stored or displayed.
One of the most basic display techniques is M-mode ultrasound, where the M
refers to motion. This in turn is based on B-mode ultrasound, where the B stands
for brightness. In B-mode, the signal returned from a single directional ultrasound
beam is displayed as a projection whose brightness is proportional to the returned
signal. Because the timing of the returned signal relates to the depth from which it

originates, the B-mode display represents a 1D projection of the tissue interfaces
along the beam.
In M-mode ultrasound, the same beam direction is repeatedly displayed from the
B-mode information. The result is a time-resolved display of the motion occurring
along this projection. This display method has commonly been used to show the
motion of heart valves, as the repeated sampling of the projection can occur very
rapidly, yielding excellent temporal resolution. However, as 2D echocardiographic
techniques have advanced, M-mode display has declined in importance, overtaken
by truly 2D and 3D display methods for most of the same information. Figure 1.2
shows a sample M-mode acquisition used to derive functional parameters over the
heart cycle.
Fig. 1.2 M-mode echocardiogram acquired over multiple cardiac cycles. The dashed lines esti-
mate systolic and diastolic parameters to derive functional indices such as stroke volume and
ejection fraction
1.2.1.2 2D Ultrasound
With the advent of transducer arrays and rapid signal processing, B-mode and
M-mode ultrasound have been almost completely supplanted by 2D imaging acqui-
sitions. The fundamental principle of 2D ultrasound is precisely the same as that
of B-mode. A pulse of acoustic energy is transmitted into the tissues, and the

10 D.R. Thedens
returned reflections or “echoes” are recorded and displayed, providing a display
of the boundaries between tissues and characteristic scatter patterns of organs.
In 2D ultrasound, a complete image is formed by sweeping over all of the trans-
ducer elements, each generating a single line of image data. The lines are assembled
into a complete 2D image for display, and the process is repeated. The transducer
may be linear, where the elements are lined up in a straight line along the transducer
to produce parallel beams and a rectangular image field of view. Another config-
uration is a curvilinear array, which uses a smaller transducer head with a convex
shape. The individual elements are fanned out over the transducer and the resulting
images have a trapezoidal field of view.
1.2.2 Echocardiography
On modern ultrasound equipment, the rate at which images are formed is fast
enough to generate real-time imaging of the beating heart. This has made ultra-
sound an extremely effective imaging modality for observing cardiac morphology
and function. When applied to cardiac imaging, ultrasound is usually referred to as
echocardiography.
For cardiac imaging, the transducer array used is convex, generating the trape-
zoidal field of view described above. This allows the transducer to be small enough
to be positioned in the limited set of locations that permit an unobstructed window
for acquiring images of the heart (in particular, without crossing air spaces in the
lungs), such as between the ribs.
Since echocardiography remains predominantly a 2D imaging technique, a com-
plete echocardiographic exam utilizes multiple imaging planes to acquire a complete
description of the anatomy and function of the heart. The long-axis view of the heart
runs parallel to the long axis of the heart (as the name suggests) and depicts the left
atrium, left ventricle (LV), septum, and posterior wall of the LV. It can also be ori-
ented to show right ventricular (RV) inflow and outflow tracts. The short-axis plane
runs perpendicular to the long axis, showing the LV in cross section. By varying
the positioning of the imaging plane, the morphology and function of the heart can
be assessed from base to apex. The four-chamber view cuts through the heart from
apex to base to show all four chambers, as well as the mitral and tricuspid valves in
a single view.
1.2.2.1 Morphologic Imaging
The capabilities of echocardiography to generate high-resolution and real-time
depictions of the cardiac anatomy from multiple vantage points are also useful for
generating quantitative assessments of many indices of cardiovascular health. The
most basic measurements relate to the size of the various chambers and outflow
tracts of the heart. In the LV, dimensions of the posterior wall and septum can be
taken either from the 2D images or directly from an M-mode projection. There are
multiple methods for estimating LV mass from echocardiograms. Short-axis dimen-
sions can be used with a simple geometric formula to produce a reasonably accurate

mass measurement, though this entails some assumptions about the geometry of
the chamber. Alternatively, two long-axis views can be used with a Simpson’s rule
derivation for a more accurate assessment of LV mass and volumes. Calculating
dimensions of the left-and right ventricular outflow tracts is also possible from the
appropriate long-and short-axis views, respectively.
With similar methods, RV volumes and thickness can be measured as well as
those of the left atrium. In general, an appropriate orientation needs to be cap-
tured over the heart cycle, followed by identification of the standard locations for
measurement of wall or chamber dimensions.
1.2.2.2 Function
The dynamic nature of echocardiographic image acquisition makes it well-suited
for studying the functional dynamics of the heart. Several functional indices can be
derived directly from the appropriate morphological measurements. Since echocar-
diography produces images of high spatial and temporal resolution, such indices are
of primary importance.
LV ejection fraction (LVEF) is the single most important index of cardiac
function and can be derived directly from measurements of end-systolic and end-
diastolic volumes measured as outlined above. Stroke volume is another important
parameter of diagnostic interest and is measured from the same parameters.
1.2.2.3 Flow (Doppler)
Doppler echocardiography utilizes the Doppler effect to ascertain the rate and direc-
tion that material is moving. Primarily, this is used to generate quantification of the
rate of blood flow, but in principle it can be applied to any moving tissue.
The Doppler effect arises when the acoustic waves of the ultrasound beam are
reflected by moving red blood cells. Stationary tissues may reflect ultrasonic waves,
but their frequency will not be affected. When a wave is reflected by a moving mate-
rial, the frequency of the wave will be increased if the material is moving toward
the transducer and decreased if it is moving away. The frequency shift depends on
the speed of sound in the moving material c, the frequency of the transmitted wave
f0, the velocity of the material v, and the angle between the beam and the motion θ.
The frequency shift f is then
f =
2f0v cos θ
c
(1.4)
Thus, once the measurement of the frequency change is taken, the velocity is
computed as
v =
fc
2f0
(1.5)
assuming the beam is parallel to the flow direction (if not, the flow rate will be
underestimated).

12 D.R. Thedens
Color Doppler imaging is the most widely used form of Doppler flow imaging
at present. Flow measurements are continuously taken by alternately pulsing and
recording the frequency shifts returned. The frequency information is mapped to a
predefined color map to display the flow information atop the anatomic data. By
convention, red is used for flow toward the transducer and blue for flow away from
the transducer, with lighter shades indicating greater velocity. In areas of turbulent
flow where the flow directions and velocities are highly variable, green is displayed.
The acquisition and display of flow information make it easy to identify cardiac
abnormalities, many of which are characterized by disturbances in flow patterns.
These include valve disease where backflow may be seen and pathologies of dias-
tolic function where flow patterns into the ventricles may be readily observed. An
example of color Doppler imaging in a subject with an atrial septal defect is shown
in Fig. 1.3.
Fig. 1.3 Doppler mode echocardiogram of the heart with color-coded overlay indicating flow
velocity and direction in this subject with an atrial septal defect
Doppler imaging is not limited to observation of blood flow. Measurements of the
velocities of myocardial tissues can also be used to estimate velocities and myocar-
dial strain rates. The velocities under consideration are much lower than those from
flowing blood, and the translational motion of the heart may cause errors in the
velocity measurements, so care must be taken in interpreting the derived values.
1.2.2.4 TTE Versus TEE
As noted previously, because of the position of the lungs relative to the heart and
the need for an air-free path to visualize the heart with ultrasound, the placement
and orientation of the ultrasound probe is limited to a few external positions on the
chest (where it is called transthoracic echocardiography or TTE). An alternative

approach is to use a special ultrasound probe that can be passed into the esophagus,
which places the probe in much closer proximity to the heart (which sits within
millimeters of the esophagus). This is known as transesophageal echocardiography
(TEE). The result is much increased reflection energy and reduced attenuation and a
corresponding improvement in image quality. Several conditions involving the left
atrium, mitral valve, pulmonary artery, and thoracic aorta are best seen on TEE.
Obviously, TEE is a much more invasive procedure than TTE, as it requires a
fasting patient and conscious sedation to position the probe in the esophagus. TEE
is therefore not a routine initial screening tool, but does provide valuable anatomic
and flow information that cannot be acquired with regular TTE.
1.2.3 Vascular/Peripheral
Because of the quantitative information about flow velocities and abnormal flow
patterns that can be generated with Doppler ultrasound, its use has expanded beyond
the heart to nearby great vessels and peripheral vessels as well, in order to assess
vascular function throughout the body.
Numerous pathologies of the aorta are routinely imaged and diagnosed by ultra-
sound, both standard TTE and TEE. Aortic aneurysms and aortic dissection are
often assessed with both TTE and TEE, and atherosclerotic plaque is also a com-
mon finding from TEE. Doppler ultrasound is becoming increasingly popular for
initial assessments of carotid artery disease. The Doppler measurements of blood
flow velocity can help identify the significance of plaques and lesions in the artery.
Doppler imaging has found uses in peripheral vessels as well, where it serves as an
inexpensive screening tool.
Vascular assessment can also be performed using intravascular ultrasound
(IVUS). IVUS utilizes a miniature ultrasound probe attached to the end of a catheter
which can then be inserted inside the lumen of a vessel. Images can then be acquired
from the inside of a blood vessel to depict its lumen and wall. In particular, this
permits direct discrimination of atherosclerotic plaque contained within the vessel
wall and quantification of both the degree of narrowing and the total plaque volume
contained therein. IVUS can also determine plaque tissue characterization, as the
calcified, fibrous, and lipid components of a lesion can be distinguished based on
their appearance on ultrasound. IVUS is most commonly applied to the coronary
vasculature, where it can be used to measure plaque burden and plan treatment prior
to angioplasty or to assess stent placement or restenosis. Because of its ability to
quantify plaque burden, IVUS is also useful for assessing efficacy of treatments for
coronary atherosclerosis.
While IVUS can provide unique information on the state of blood vessel lumen
and walls, it is an invasive technique compared to the other tomographic imaging
methods discussed here, though it does not require a contrast agent as conventional
angiography does. Imaging is also limited to vessels large enough to accommodate
the probe, and positioning within large vessels may result in oblique cross sections

14 D.R. Thedens
due to angulation of the probe. Nevertheless, IVUS has proven highly valuable in
understanding the characteristics and development of atherosclerotic lesions.
1.3 Computed Tomography (CT)
CT was the first of the tomographic imaging technologies to permit generation of
images representing cross-sectional “slices” of the internal anatomy. The first CT
scanner was installed in 1972 and the initial application was in brain imaging. The
first images required a 4–5 min scan time and produced images with an 80×80
pixel matrix. Subsequent advances in x-ray tubes, detectors, and computer hardware
have improved on these characteristics by many orders of magnitude. CT scanners
are now capable of acquiring and reconstructing large 3D data sets in a few sec-
onds, making it possible to visualize minute structures and dynamic processes with
exquisite clarity.
1.3.1 Principles of CT
This section provides an overview of the principles of CT imaging. For a more
comprehensive treatment, see the relevant chapters of Ref. [3]. CT is based on the
principles of x-ray imaging. X-rays are generated in a vacuum tube by firing elec-
trons at a target (the anode) which produces a beam of electromagnetic radiation
in the x-ray spectrum. The beam is directed toward the body, and a detector on
the opposite side (which may be film or some type of solid-state or digital device)
records the amount of x-ray energy that passed through the body. In essence, the
detector serves to measure the attenuation experienced by the x-ray beam as it passes
through the body and creates a 2D projection image of the 3D anatomy. Contrast
between body tissues is developed because high-density tissues such as bone will
absorb greater amounts of energy than low-density soft tissues.
A fundamental limitation of x-ray imaging is the projective nature of the result-
ing image, meaning that structures in the third “depth” dimension are overlaid on
each other, requiring multiple views to elucidate the arrangement of structures. To
overcome this limitation, tomography was developed in the early 1900s, exploiting
principles of projective geometry. The x-ray source and detector are simultaneously
rotated around a central focus point as the x-rays are generated. Structures at the
focal point remain in focus throughout this motion, while structures away from the
focus will be blurred out and appear as noise. The result is an image showing only
the internal structures at this focal point. This form of tomography can be considered
to be a precursor of modern CT imaging, which has almost entirely supplanted it.
Similarly, the mathematical underpinnings needed for CT image formation have
a long history, originating in the work of Joseph Radon in a paper published in 1917.
The Radon transform and its inverse describe the relationship between an unknown
object and a set of line integrals or projections through the object. The remaining

development necessary for modern CT imaging to become feasible was the appli-
cation of the digital computer to perform the numerical computations required to
generate a tomographic image from a set of angular projections. CT imaging is
thus one of the first imaging modalities made possible by advances in computer
technology.
1.3.1.1 Basic CT
The basic process of image formation in CT imaging is the collection of a set of
projections taken at multiple angles which can be subsequently reconstructed into
a 2D imaging slice through the same region of the body. In conventional CT imag-
ing, a fan-beam geometry is commonly used to generate the measurements. The
x-ray generator can be considered to be a point source, and a set of diverging rays
are emitted and pass through the body. On the opposite side of the scanner bore,
detectors are arranged to measure the x-ray attenuation over the entire beam.
The attenuation of the beam follows the relationship
Idetected = Itransmittede−μx
(1.6)
where I represents an x-ray intensity and x is the thickness. The transmitted intensity
is also measured at the detector as a reference value. The resulting attenuation μ
measured will be an average over the path of the x-ray beam at each location. The
attenuation coefficient can then be computed as
μt = loge(Itransmitted/Idetected) (1.7)
Since the transmitted intensity is available at the detector, this relationship is
inherently normalized for the intensity of the beam, leaving only dependence on the
attenuation characteristics of the body tissues. As a result of this transformation,
the intensity values displayed on a CT image have a physical meaning in terms of
the attenuation coefficients at each location in the generated slice. For computation
and display, the attenuation coefficient is further normalized to the Hounsfield scale
(named for one of the Nobel-winning inventors of CT), measured in Hounsfield
Units (HU), which relates the attenuation to that of distilled water as
HU =
μtissue − μwater
μwater
× 1000 (1.8)
On this scale, pure water has an attenuation of 0 HU, while air has an attenuation
of −1000 HU. The use of this scale permits direct identification of tissue types in
images based on their measured value of HU and known attenuation characteristics
and is useful for diagnosis and subsequent image processing and visualization.
Formation of a complete image requires recording attenuation measurements
over a full 180
◦
range of angles. Numerous techniques have been developed over the
years as the sophistication of detectors and control of the hardware have improved.
Early “first-generation” scanners utilized a single detector and required a sequence

16 D.R. Thedens
of translate–rotate motions. As the number of detectors that could be incorporated
into the scanner increased, this gave way to systems with a few hundred detec-
tors that required only rotational motion, with source and detector array rotated on
opposite sides of the patient. Subsequent generations of scanners utilize even more
detectors, completely encircling the bore of the scanner (a few thousand in total) and
requiring motion of the x-ray source only, which allows for faster scan times. The
most recent advances in imaging include helical CT, whereby data can be acquired
continuously while the table moves through the bore, rather than needing to stop
and start for each set of slices. With this combination of technologies, imaging in
the span of a single breath hold became possible.
1.3.1.2 Multidetector CT
The present state of the art in CT scanning is focused on the use of multiple detector
arrays to further increase the speed and efficiency of the acquisition. This arrange-
ment is known as multidetector CT (MDCT). MDCT retains the ring of detectors
surrounding the bore of the scanner, but instead of single detectors at each location,
an array of densely packed detectors is assembled (Fig. 1.4). Thus, for a station-
ary location within the scanner bore, a number of images equal to the number of
elements in the array can be recorded and reconstructed. The dense packing of the
detectors also means that the slice thickness achievable is now dependent on the size
of the detector that can be constructed, rather than on the width of the x-ray beam
produced. Alternatively, the data from multiple elements of the array may be com-
bined together to generate thicker slices of higher quality than would be generated
by the single elements of the detector array.
Fig. 1.4 In a multidetector
CT system, the collimated
beam from an x-ray source
passes through the material
being imaged and an array of
detectors records the
transmitted energy at the
different projection angles in
parallel
As of this writing, MDCT scanners with arrays capable of generating 64 slices
are becoming commonplace, and even larger arrays are appearing on the market,
with 256-slice arrays now available from multiple manufacturers. Because of their
highly parallel nature, 64-slice and higher scanners can take a complete 4D data set
(3D spatial information with 0.5 mm acquired slice thickness and on the order of
150 ms reconstructed temporal resolution) over the heart in a short breath hold.

1.3.2 Cardiac CT
The capability to rapidly generate such comprehensive visualizations of the beat-
ing heart has vaulted MDCT to a premier position for assessment of cardiac and
coronary anatomy.
1.3.2.1 Coronary Arteries
A primary use of cardiac CT is to assess the coronary vasculature for stenoses and
calcifications associated with heart disease [4]. The current generation of 64-slice
MDCT scanners now has sufficient spatial and temporal resolution to permit accu-
rate assessment of coronary artery stenosis. High spatial resolution is needed to
identify coronary artery disease in at least the major coronary vessels. High tempo-
ral resolution is needed to be able to acquire this level of resolution in a short breath
hold as well as to accommodate high heart rates and arrhythmias.
The resolution capabilities of 64-slice and higher MDCT scanners are approach-
ing the resolution of conventional (invasive) angiography. MDCT can now realize
resolutions on the order of 0.4 mm slices, compared to a nominal resolution of
about 0.2 mm for conventional angiography. As a result, MDCT is increasingly
used to assess the severity of disease and to reduce the need for conventional
angiography in patients who do not show severe stenosis and can thus rule out
coronary artery disease. Figure 1.5 shows an example of MDCT used for detecting
coronary artery stenosis. MDCT also has the advantage of the possibility of deter-
mining tissue characteristics of stenoses, such as calcification. Three-dimensional
reconstructions can aid in localization of lesions and planning of interventions.
Fig. 1.5 MDCT of the left coronary artery in a patient with an occlusion in this vessel. The left
panel displays the artery in a single plane view, while the right panel shows a 3D reconstruction of
the heart and coronary vessels

18 D.R. Thedens
Nevertheless, conventional angiography remains the gold standard measure of coro-
nary status, particularly for collateral vessels. This may continue to be revisited as
MDCT continues to improve in resolution and scan time.
A secondary use of MDCT in coronary arteries is coronary artery calcifica-
tion scoring. Calcifications in the coronary arteries are readily visualized on MDCT
because of their high density, and many studies have shown a high degree of cor-
relation between calcium scoring and overall plaque burden, which in turn may
predict the risk of future cardiac events. Again, the use of 64-slice scanners makes
the acquired resolution detailed enough to eliminate many of the partial voluming
effects and other limitations of previous generations of scanners and yield greater
accuracy in this assessment, though the risk of false-positive results from these
measures has not been eliminated.
1.3.2.2 Aorta
MDCT with intravenous iodinated contrast is also widely used for detecting and
assessing problems in the thoracic and abdominal aorta [5]. The high resolution and
volume coverage of MDCT can serve as the basis for 3D visualization of the lumen
of the aorta.
Aortic dissection is a tear in the wall of the aorta, which permits blood to flow
between the layers of the wall and further forces them apart, with the risk that the
aorta itself will rupture with fatal consequences. MDCT (along with MRI) detects
dissection with a high degree of sensitivity and specificity. Though MRI remains
the gold standard for this condition, the more rapid scan time and higher resolution
of MDCT may be preferable in many instances. Similarly, aortic aneurysms are
well visualized and followed on MDCT [6]. Detection of aneurysms again relies on
contrast-based examination of the vessel.
1.3.2.3 Cardiac Function
The rapid scan times, with temporal resolution approaching 60 ms, along with the
resolution of MDCT in the heart, has generated interest in its use for cardiac function
assessment [7]. Presently, MRI is the primary standard for measuring such indices as
left ventricular (LV) ejection fraction, end-diastolic and end-systolic volumes, and
LV mass. The submillimeter slice thickness possible with 64-slice CT has brought
its accuracy for these measures to a sufficient level such that MDCT is making
inroads for LV function assessment because of its rapid scan times.
The present limitations of MDCT for cardiac function are the temporal resolu-
tion (which is limited by the rotation speed of the scanner gantry) and the short
scan times, where there may be variability in these parameters from heartbeat to
heartbeat. Further advances in CT such as larger numbers of slices and dual source
systems will likely continue to close this gap.
A dual-source CT scanner is equipped with two x-ray sources and two corre-
sponding detectors, oriented at 90
◦
to each other. The two sources and detectors
can operate simultaneously, acquiring twice as much data at a time compared to a

single-source system. This permits complete image acquisitions in half the time, a
particular advantage for cardiac imaging as it improves the temporal resolution and
reduces the required time for a breath hold.
1.3.3 Pulmonary CT
The high resolution and rapid scan times now available with MDCT have opened
up its use for a variety of indications in the lungs. In order to prevent artifacts
from motion in the lungs, CT imaging of the lungs requires a breath-holding
protocol. Previously, the length of the scan required to generate the needed reso-
lution would require breath-holding durations that were prohibitively long for many
patients with lung disease. With MDCT now permitting high-resolution imaging
within seconds, such acquisitions now become feasible, and MDCT has become the
modality of choice for numerous clinical and research questions. The resolution of
MDCT imaging is the highest among tomographic imaging methods suitable for
lung imaging.
1.3.3.1 Parenchyma
CT is among the only imaging modalities capable of generating high-resolution
tomographic imaging of the lung parenchyma, as the tissue densities are unsuitable
for imaging with ultrasound or MRI. As a result, CT is the preferred method of
imaging for assessment of lung nodules and staging of treatment. Additionally, CT
can be used to quantify the extent and severity of chronic obstructive pulmonary
disease (COPD) and emphysema based on areas of reduced attenuation that result
from these conditions [8].
The continuous coverage generated by CT imaging coupled with the potential
for very thin slice acquisitions permit CT acquisitions to be arbitrarily reconstructed
into 3D volumes to generate 3D models of the airways. These features are used in
applications such as CT bronchography and virtual bronchoscopy. These methods
are enhanced by the use of 3D rendering techniques and visualizations that allow
both qualitative and quantitative assessments of airways. As the resolution of MDCT
continues to improve, smaller and smaller airways can be visualized and measured
for ever-growing understanding of the normal and diseased lung.
The use of inhaled xenon gas as a CT contrast agent has yielded improvements
in measurements of regional pulmonary ventilation [10]. Xenon has a high atomic
number and is thus much more radiodense in proportion to its concentration com-
pared to air or soft tissues; it therefore yields high contrast against such tissues.
Xenon-enhanced CT (Xe-CT) involves inhaling and exhaling the gas during a time
series of imaging acquisitions. Local and regional ventilation time constants can
then be derived from the rate of the gas movement. While presently Xe-CT is pri-
marily a research tool, it provides unique and valuable information on lung structure
and function.

20 D.R. Thedens
1.3.3.2 Pulmonary Angiography
Assessment of the pulmonary vasculature has also become a routine use of MDCT in
the lungs. As with other modalities, pulmonary angiography generally involves the
introduction of a contrast agent (an iodinated agent in the case of CT). The primary
use of pulmonary angiography is in the identification of pulmonary embolism [9].
Figure 1.6 shows a sample MDCT acquisition depicting a “saddle” embolism in the
left and right pulmonary arteries.
Fig. 1.6 MDCT image
through the pulmonary artery
demonstrating a pulmonary
embolism
1.4 Magnetic Resonance Imaging (MRI)
1.4.1 Principles of MRI
This section describes the basic principles behind the formation of images with
MRI. Detailed descriptions of the physics and instrumentation are provided in Refs.
[11, 12].
Magnetic resonance imaging (MRI) relies on the phenomenon of nuclear mag-
netic resonance to generate image contrast. The hydrogen atom (along with other
species having an odd number of protons or neutrons, such as sodium and phospho-
rous) possesses a spin angular momentum. The single proton of the hydrogen atom
(often referred to in this context as a spin) is by far the most abundant and thus is
considered in the vast majority of imaging applications. Most importantly, for the
purposes of imaging, the spins will give rise to a magnetic moment and will act
like microscopic bar magnets. As a result, when the protons are placed in a strong
static magnetic field, at equilibrium they tend to line up in the same direction as the
external field. The net effect of all the spins lined up in this way generates a small
but measurable magnetization along the longitudinal direction of the large external
field. The magnitude of this magnetization increases as the strength of the external
field is increased.

1.4.1.1 Signal Generation
By itself, the magnetization does not give much useful information about the distri-
bution of the protons within the object. The application of a second small (relative
to the primary strong) magnetic field oscillating in the radiofrequency (RF) range
sets up a resonance condition and will perturb the spins away from their equilib-
rium state, “tilting” them away from their alignment with the main field into the
transverse plane. Much like a gyroscope, this will excite the spins, causing them
(and their magnetic fields) to precess about the direction of the main field, and the
rate at which the spins precess is directly proportional to the strength of the main
magnetic field. Figure 1.7 shows the relationship between the two magnetic fields
and the resulting perturbation of the magnetization vector. A fundamental principle
of electromagnetics is that a time-varying magnetic field can induce an electric cur-
rent in an appropriately placed coil of wire, generating a signal that can measure the
distribution of the spins within the object. Since the rate of precession depends on
the magnetic field strength, slightly varying the field strength across the bore of the
magnetic with gradient fields yields a spatially varying rate of precession. When the
RF field is removed, the spins begin to return toward their equilibrium state, aligned
with the strong static magnetic field.
Fig. 1.7 A rotating RF
magnetic field with frequency
ω 0 is applied perpendicular
to the direction of the main
magnetic field. This causes
the aligned spins to tip away
from the main field direction
and precess at the same
frequency ω 0, producing a
detectable signal from the
spins
The rate of return of spins to their equilibrium state is governed by two time con-
stants intrinsic to different tissue types, T1 and T2. T2 determines how long it will
take for the signal generated by the “tipped” spins to decay away. T1 measures the
amount of time it takes for the spins to completely return to their equilibrium align-
ment with the main magnetic field. Because of this signal decay, an MR imaging
experiment generally must consist of several cycles of signal generation followed
by signal measurement or acquisition.
1.4.1.2 General Techniques and Contrast Mechanisms
The signal measured from a tissue will thus depend on its density of protons as
well as its T1 and T2 relaxation parameters. Motion and flow also contribute to the
final signal generated. The remarkable ability of MRI to generate a wide variety
of tissue contrast arises from the fact that the imaging experiment can be designed

22 D.R. Thedens
to vary the relative weight of each of these parameters in the measured signal. For
example, muscle and fat have very different T1 and T2 parameters, and by varying
the timing of the applied RF excitation pulses, maximum contrast between the two
can be achieved. Other strategies may enhance or suppress flowing blood compared
to stationary tissues.
1.4.1.3 Morphology
The most basic use of cardiac MRI is to depict the structure or morphology of
the heart. Two general classes of imaging techniques are widely used for cardiac
imaging, commonly referred to as black-blood and bright-blood techniques.
Black-Blood Imaging
Black-blood images are produced by T2-weighted spin-echo (SE) imaging
sequences [13], in which two RF excitations (an excitation pulse and an inversion
pulse) are applied to the imaged volume. After the excitation pulse, the excited
spins begin to lose coherence due to slight variations in their resonant frequencies,
resulting in a rapid loss of overall signal.
The inversion pulse “flips” the magnetization about one of the axes, permitting
these spins to regain their coherence and generate an echo when the signal has been
restored. When the two pulses are separated by a sufficient interval, flowing blood
experiences only one of these pulses and thus does not produce a restored signal
echo, leaving a flow void in the chambers of the heart. The timing of the two RF
pulses sets the echo time (TE) at which the signal refocuses (and data are acquired)
and determines the precise signal and contrast features of the image. For black-blood
imaging, a TE of at least 20 ms is usually used. A longer TE yields greater con-
trast based on T2 characteristics of the tissues, which may be useful to identify such
lesions as acute myocardial infarction or myocardial scar. This comes at the expense
of reduced overall signal due to signal decay. Standard SE sequences show excel-
lent contrast among myocardium (medium intensity), epicardial fat (high intensity),
and flowing blood (low intensity). The signal void created by SE sequences gener-
ates images with especially good contrast in endocardial regions, valves, and vessel
walls.
The main limitation of standard SE sequences is the acquisition time required in a
cardiac-triggered exam, which results in poor temporal resolution and the prospect
of significant respiratory motion artifact. Fast SE (FSE) sequences overcome this
limitation by applying multiple inversion pulses and additional signal readouts dur-
ing a single cardiac cycle. Speedups of an order of magnitude are possible in this
way. However, the longer readout times degrade the image contrast due to the more
complex dependence on relaxation times.
The currently preferred black-blood technique for imaging cardiac morphology
is a T2-weighted inversion recovery (IR) pulse sequence. This sequence applies
additional RF excitation pulses to effectively null the signal from blood (and possi-
bly fat as well) based on its T1 relaxation parameters. This is usually followed by

a FSE sequence that can be acquired in 15–20 heartbeats, suitable for a breath-held
acquisition and yielding a robust black-blood sequence with T2 contrast.
Bright-Blood Imaging
Bright-blood images originate from gradient echo (GRE) imaging sequences which
only use a single RF excitation, relying on the gradient hardware instead of an
inversion pulse to refocus the signal for data acquisition. Much shorter TE times
(1–10 ms) are used, and the excitation and data readouts can be repeated more fre-
quently (every 10–20 ms). Because the blood need only experience the single RF
pulse to generate a signal, it appears brighter than myocardium on GRE acquisitions.
The short TE between excitation and data readout enhances this effect since there
is less time for signal decay due to relaxation. Additional flow-compensation pulses
can also be applied to further enhance blood signal and improve contrast with nearby
myocardium. As with FSE imaging, the fastest imaging sequences utilize multiple
excitations and data readouts over an extended interval (80 ms is a typical duration)
synchronized to the cardiac cycle to generate images that can be acquired within a
breath-holding interval. Contrast between blood and myocardium is generally not
as good as with SE imaging, as varying flow profiles may result in heterogeneous
blood pool.
The availability of faster gradient hardware has seen a resurgence of techniques
based on steady-state free precession (SSFP) [14]. SSFP maximizes the use of sig-
nal from blood by applying rapid excitations repeated at very short intervals. The
resulting contrast is a function of relaxation parameters as T1/T2. The short rep-
etition times greatly reduce flow effects and show a more homogeneous depiction
of myocardial blood pool, which in turn improves contrast with myocardium and
visualization of papillary muscles. Rapid excitations also permit better temporal
resolution, or the time savings can be traded off for higher resolution at the same
time resolution. As state-of-the-art MR gradient hardware proliferates, SSFP will
likely become even more common.
The rapid repetition of readouts in both GRE and SSFP means that several images
at the same location can be taken at different time points within the heart cycle.
Alternatively, the imaging time can be used to acquire multiple slices at a reduced
temporal resolution. Using segmented acquisitions, a multi-slice multi-phase view
of the cardiac morphology can be acquired within a single breath hold of 15–20
heartbeats.
1.4.1.4 Function
Many of the techniques mentioned above for imaging of cardiac morphology,
including both black-blood and bright-blood imaging, are also suitable for measur-
ing cardiac function indices. Compared to other modalities, MRI has the advantage
that completely arbitrary image orientations can be chosen, guaranteeing that true
long-axis or short-axis views serve as the basis for quantitative measurements.

24 D.R. Thedens
Fig. 1.8 Cardiac MRI two-chamber views of the left atrium and left ventricle
Figure 1.8 shows an example of a two-chamber view of the left ventricle. The avail-
ability of 3D information in the form of multiple parallel slices eliminates the need
for any geometric assumptions about ventricular anatomy when estimating masses
and volumes, a significant advantage over x-ray and ultrasound.
Bright-blood GRE imaging is more commonly used for evaluation of ventricu-
lar function. The shorter acquisition time permits a greater number of slices to be
acquired during the cardiac cycle, which can be used for higher temporal resolution
(more frames per cycle) or for a greater volume coverage (more slice locations).
The acquisition of images at multiple phases of the cardiac cycle is known as cine
MRI (example shown in Fig. 1.9). With present system hardware, a complete multi-
slice multi-phase cine data set suitable for quantitative analysis can be acquired in
a single breath-hold interval. The limiting factor with standard GRE imaging is the
contrast between medium-intensity myocardium and the bright-blood pool. Areas
of slower flowing blood will demonstrate reduced intensity, making delineation of
the endocardial contours difficult.
Fig. 1.9 Sequence of cardiac MRI short-axis images of the left ventricle. Slices are shown from
base to apex (left to right panels) in the diastolic phase of the cardiac cycle
The recent advances in SSFP imaging cited above may solve this problem to
some degree with more robust contrast. The faster repetition time used in SSFP

also increases the frame rates possible in a cine study. With state-of-the-art gradient
hardware, truly 3D cine MRI with no gaps between slices is now possible within a
single breath-hold interval.
Improving gradient and computing hardware has now made real-time imaging
feasible for functional imaging. Rates of 16 frames per second or more can be
continuously obtained, much like x-ray fluoroscopy. The scan plane can be mod-
ified directly on the real-time images, dramatically reducing the time required for
“scout” scans to find the proper short-axis orientation. At such rates, cardiac gat-
ing and breath holding are unnecessary, which permits imaging of patients with
arrhythmias. Presently, spatial resolution of real-time studies remains comparatively
limited, but a number of ongoing developments in image reconstruction techniques
are improving this. Two general strategies exploit the widespread use of multiple
receiver coils. Simultaneous acquisition of spatial harmonics (SMASH)and sensi-
tivity encoding (SENSE) use the spatially varying response of a group of coils as an
additional means of spatial encoding to reduce the time needed to acquire a given
resolution image. Other techniques analyze the temporal dimension of the acquisi-
tion to reduce the acquisition of redundant information and enhance either temporal
or spatial resolution.
Each of these forms of cine and real-time MRI data is useful for computing
several global measures of cardiac function. Accurate and reproducible quantita-
tive measurements of ventricular volumes at both systole and diastole, masses, and
ejection fraction (difference between the diastolic and systolic ventricular volumes)
are all computable with multi-slice or volume data sets. In each case, myocardial
border identification is necessary to extract quantitative results. Compared to x-ray
and ultrasound, MRI also accurately depicts epicardial borders, again eliminating
the geometric assumptions that often must be made in competing modalities. As a
result, regional myocardial function assessments can also be made with cine tech-
niques. This may be done subjectively, viewing cine or real-time “loops,” or through
quantitative measurements of regional wall thickness and strain.
Regional measurements of 3D strain are possible using myocardial tagging. This
imaging method excites myocardium with a pattern of lines or grids whose motion
can then be tracked over the heart cycle, providing a precise depiction of the defor-
mations occurring within the myocardial tissues. Analysis of these deformations in
short- and long-axis views gives 3D strain measurements useful for determining
local myocardial function. A promising rapid technique is harmonic phase (HARP)
imaging, which has potential as a real-time technique.
1.4.1.5 Perfusion/Ischemia
Another important indicator that can be assessed by MRI is regional blood flow
(or perfusion) in the myocardium [15]. This may indicate areas of damage to
myocardium from a cardiac event or insufficient blood flow resulting from a signif-
icant arterial stenosis. Determination of blood flow within the myocardium depends
on the use of contrast agents (usually gadolinium-based) that change the relaxation
characteristics of blood, particularly the T1 relaxation time. Gadolinium causes

26 D.R. Thedens
a considerable shortening of the T1 relaxation time, meaning that magnetization
returns to equilibrium much more rapidly. When RF excitation pulses are applied
in rapid succession, tissues with short T1 relaxation will still have time to recover
and generate greater signal for subsequent excitations. Longer T1 relaxation times
means that little magnetization has returned to the equilibrium state, so later exci-
tations result in much less signal. Appropriate timing of a pair of RF pulses can
maximize the signal difference between two tissues with known T1 relaxation times.
Perfusion is mostly measured during the “first pass” into the myocardium after
injection of the contrast agent. Areas of myocardium with adequate blood flow will
have enhanced intensity from the shortened T1 of the inflowing blood. Perfusion
deficits will not receive this material and remain at lower intensity. The time of the
imaging window is limited as contrast material may soon begin to diffuse from nor-
mal to deficit regions, and the contrast agent will recirculate with the blood within
15 s. Hence, rapid GRE sequences are used to image quickly and permit multiple
slices to be obtained over a volume. T1 contrast is maximized by applying an RF
“preparation” pulse that initially excites or saturates all of the blood and tissues.
After a delay time that causes contrast-enhanced material to return toward equilib-
rium while the longer T1 tissues recover much less magnetization to yield strong
T1 contrast, a standard fast GRE imaging sequence is applied. The result is bright
signal in normal tissue and low-intensity regions of perfusion deficit. Acquisition
of several time frames during this process permits quantitative measurements of the
severity of these perfusion abnormalities. Further myocardial tissue characteriza-
tion is possible using gadolinium contrast agents by waiting an extended duration
(20 min or more) before imaging. Gadolinium contrast will eventually move to the
extracellular space and accumulate more in areas of non-viable myocardium, result-
ing in enhanced signal in these areas on T1-weighted images compared to normal
tissue.
1.4.2 MR Angiography
In addition to imaging of the heart, MRI has also been widely applied to imaging
vessels throughout the body [16]. Its advantages over conventional x-ray angiog-
raphy go beyond the fact that it is much less invasive. MRI can also collect true
3D data, permitting arbitrary selection of views and slices in post-processing to
optimize the visualization of vessels. This is especially helpful in complex vascular
trees where tracing the vessel of interest may be difficult. Contrast for MR angiogra-
phy can be developed in two ways. Pulse sequences may exploit the different signal
properties of flowing and stationary tissues to produce images. Other sequences rely
on the relaxation characteristics of arterial and venous blood, usually enhanced by
T1-shortening contrast agents as described for myocardial perfusion. In both cases,
the goal is to generate images of the vessel lumen suitable to detect and evaluate
stenoses.

Two flow-based imaging techniques are in common use for MR angiography and
both effectively produce “bright-blood” images of the vessel lumen. Phase-contrast
(PC) imaging takes advantage of the fact that flowing blood will move during the
data acquisition readout. Since spatial information is encoded by a spatially vary-
ing magnetic field gradient, flowing spins experience a changing magnetic field as
they move, resulting in a phase change in their signal compared to stationary tis-
sues. By applying an appropriate encoding gradient pattern prior to imaging, flowing
blood can be selectively viewed. PC imaging can also quantitatively measure flow
velocities. Time-of-flight (TOF) imaging uses the continuous replacement of flow-
ing blood in the imaged slice to differentiate it from static tissue. Rapid repetition
of excitation pulses covering the imaged slice saturates and eventually eliminates
signal from stationary material because there is not enough time to regain any equi-
librium magnetization. Flowing blood retains signal since fresh unsaturated blood
is constantly flowing into the slice to be excited and flows away again before satu-
ration can be complete. The result produces high signal from flowing blood against
the low intensity of background structures.
Reliance on flow for image contrast may introduce artifacts where flow patterns
are not ideal. Such anomalies will affect both PC and TOF sequences. Areas of
slow flow may have reduced signal, either due to reduced phase changes for PC or
saturation in TOF. Complex flow patterns and turbulence can also cause reduced
intensities within the vessel lumen in both cases. The consequences could include
stenoses that are overestimated or a false appearance of an occlusion of the vessel.
The limitations of flow-based angiography have made flow-independent tech-
niques more prevalent. It is possible to create high-contrast angiographic images
using only the intrinsic T1 and T2 relaxation characteristics of blood through a vari-
ety of “prepared contrast” pulses that saturate or suppress one or more background
tissues. However, injectable contrast agents such as those based on gadolinium com-
pounds have proven to be safe and well tolerated and are widely available. These
contrast agents dramatically reduce the T1 relaxation time of blood and greatly
enhance its signal on TOF images. Much of MR angiography is now dominated
by contrast-agent-based protocols.
Once again, the main limiting factor in contrast studies is the time before the
contrast agent leaks outside the blood vessels and begins to enhance the signal in
tissues other than blood. Successful contrast angiography therefore requires careful
timing of contrast injection and image acquisition and a rapid acquisition technique
to minimize artifacts due to contrast dispersion and respiratory motion. Fast 3D
GRE imaging is most commonly used to acquire T1-based contrast to yield bright
contrast-enhanced blood pool. Subtraction of a non-contrast-enhanced volume may
also be used to further suppress background structures. A variety of strategies have
been employed to reduce the imaging time to acquire a 3D data set even further
and assure accurate timing of the acquisition. Partial acquisition methods which
acquire 60–75% of a full data set and synthesize the rest based on mathematical
assumptions can help reduce imaging times. More extreme versions of this have
been applied to radial sampling patterns to reduce acquisition time even further,
trading the shortened time for some increased and coherent background noise.

28 D.R. Thedens
The timing of the acquisition relative to the injection of contrast agent is also cru-
cial. If the data acquisition occurs too early, signal will not yet be enhanced, while
a late acquisition will show poor contrast because of heightened signal from other
tissues or veins. For many applications, a fixed time delay based on previous expe-
rience may be sufficient, although increased doses of contrast often accompany this
technique to increase the window of enhancement. A much smaller dose of contrast
may be given and tracked with a sequence of rapid 2D images that may be used
to pinpoint the transit time prior to a full 3D acquisition. Automatic monitoring of
the signal at a predefined location upstream from the desired location has also been
implemented. The use of real-time imaging to monitor contrast passage is another
possibility.
The limited volume imaging time available because of the dispersion of con-
trast agent into other tissues is currently being addressed. New intravascular contrast
agents that do not leak into tissues during the course of a typical MR exam are being
perfected by a number of researchers. As a result, their T1 shortening properties can
be utilized for longer or multiple exams without the enhancement of background tis-
sues. MR angiograms in higher resolution 3D or over the whole body then become
possible. The longer persistence in the blood pool does mean that both arteries and
veins will be displayed for longer 3D scan durations. Some means of separating the
two may be needed for diagnostic use of such images.
Coronary artery imaging may be a particular beneficiary of such contrast agents,
as the necessity of high resolution, 3D coverage, and motion correction requires
longer scan times than are feasible with standard contrast material. The flow and
saturation effects that often compromise 3D techniques are also improved with
such contrast agents. Perfection of a minimally invasive coronary MR imaging
exam is of particular interest because of the number of highly invasive x-ray
angiography procedures that are performed that show no clinically significant
disease.
1.4.3 Pulmonary MRI: Emerging Techniques
As noted previously, MRI of the lungs has generally been limited to contrast-
enhanced angiography of the pulmonary vasculature, as the lung parenchyma
has low proton density and significant susceptibility effects from the air–tissue
interfaces.
Hyperpolarized gas MRI has emerged in recent years as a promising new
approach to imaging of lung structure and function [17]. Rather than using the
inherently low proton density of lung tissue as a basis for imaging, an inhaled MR-
sensitive gas (such as 3He) is used as a contrast agent. To overcome the extremely
low density of spins, the gas first undergoes the process of hyperpolarization, which
aligns the spins and produces the magnetization that would normally be accom-
plished with the main magnetic field of the scanner. The normal level of polarization

(the number of excess spins preferentially aligned with the magnetic field and thus
capable of generating a signal) for a clinical strength magnet is on the order of
5–10 parts per million (0.0005%); the processed gas typically reaches polariza-
tion levels of 25%, an increase of a factor of 50,000. This more than compensates
for the density differences and permits generation of high-quality (high SNR)
images.
An important limitation of hyperpolarized gas imaging is that (unlike standard
proton imaging), the polarization is not renewed during the scan. That is, after
an excitation, the magnetization generated returns to its thermal equilibrium state
rather than the hyperpolarized state (which is effectively zero). As a result, imaging
pulse sequences for hyperpolarized media must carefully manage the polarization
to generate images of sufficient quality.
Hyperpolarized 3He imaging can provide detailed maps of lung ventilation, as
shown in Fig. 1.10. Because only the helium gas produces signal in such images,
the signal intensity in the acquired images is related directly to the distribution of
the gas within the lungs. In addition to ventilation information, additional imag-
ing techniques such as diffusion imaging can yield more structural and functional
information. Diffusion imaging produces a map of the apparent diffusion coefficient
(ADC). This parameter will be affected by the size of the lung structures where the
gas is located, with small structures restricting the range over which the gas may dif-
fuse. Increases in ADC can indicate a local loss of structural integrity and are useful
for assessing conditions such as emphysema. Figure 1.10 also shows an example of
an ADC map from a volunteer corresponding to the ventilation scan in the middle
panel.
As with Xe-CT, hyperpolarized 3He imaging remains a research tool that
does provide unique information that cannot otherwise be obtained noninvasively.
Hyperpolarized xenon gas can also be used for MRI to generate similar informa-
tion. Future work may see an emerging role for these gas-based contrast agents for
noninvasive diagnosis and treatment assessment.
Fig. 1.10 Depiction of lung ventilation using hyperpolarized 3He MRI. Ventilation images are
shown for two slice thicknesses (left and center panels), along with a corresponding map of appar-
ent diffusion coefficient (ADC), which relates to dimensions of the lung microstructure and is
sensitive to disease states

Other documents randomly have
different content

Hänen sydämensä kivut olivat kuitenkin niin kauheat, että tunteet
lopultakin saivat voiton. Kolme päivää siitä, jolloin hän tietoisesti näki
rakkautensa, hän sai pistetyksi säkeensä kuoreen ja kirjoitetuksi
Gösta Berlingin nimen siihen. Niitä ei lähetetty kumminkaan.
Ennenkuin löysi sopivan kirjeenviejän hän sai kuulla Gösta Berlingistä
semmoista uutta, että tajusi, että oli myöhäistä enää voittaa häntä
takaisin.
Mutta hänen elämänsä suruksi jäi, ettei hän ollut lähettänyt runoa
ajoissa, jolloin vielä olisi ehkä voinut hänet saada.
Kaikki hänen tuskansa kiertyi ja solmiutui tähän: Kunpa en silloin
olisi vitkastellut niin kauan, jos en olisi viivytellyt niin monta päivää!
Elämän onnen tahi ainakin elämän todellisuuden olisivat ne
hankkineet hänelle, nuo kirjoitetut sanat. Hän oli varma, että ne
olisivat tuoneet Göstan takaisin hänelle.
Suru teki kuitenkin hänelle saman palveluksen kuin rakkaus. Se
teki hänet kokonaiseksi ihmiseksi, kykeneväksi antautumaan niin
hyvään kuin pahaan. Kuohuvat tunteet virtasivat vapaasti hänen
sielussaan, itsetarkastelun jään ja hyyn niitä estämättä. Niinpä hän
tuli, huolimatta rumuudestaan, hyvin rakastetuksi.
Sanotaan kuitenkin, ettei hän koskaan Gösta Berlingiä unohtanut.
Hän suri häntä niinkuin surraan hukattua elämää.
Ja hänen runoraukkansa, joita siihen aikaan hyvin paljon luettiin,
ovat kauan sitten joutuneet unhoon. Katsellessani niitä nyt ne
kumminkin tuntuvat oudon liikuttavilta: ne on piirretty tiheällä, sirolla
käsialalla, mutta paperi on jo kellastunut ja muste haalistunut. Koko
elämän kaipuu on solmittu noihin poloisiin sanoihin, ja minä kopioin

niitä mystillisestä tunteesta vavisten, ikäänkuin niissä piilisi salaisia
voimia.
Pyydän teitä lukemaan ne ja ajattelemaan niitä. Ken tietää mikä
mahti niillä olisi ollut, jos ne olisi lähetetty? Ovathan ne kyllin
kiihkeitä ollakseen todellisen tunteen todistuskappaleita. Ehkä ne
olisivat voineet tuoda Göstan takaisin hänelle.
Ne ovat liikuttavia ja helliä avuttoman muodottominakin. Kenpä
toivoisi niitä toisenlaisiksi. Kenpä tahtoisi ne riimin ja mitan kahleihin
kurotuiksi, ja kumminkin on niin haikeaa ajatella, että kerran ehkä
juuri tuo epätäydellisyys esti häntä lähettämästä niitä ajoissa.
Pyydän lukemaan ne ja rakastamaan niitä. Suuren vaivan
ahdistama ihminen on ne kirjoittanut.
Laps, olet lempinyt; koskaan et
nyt rakkauden riemua maistaa saa.
Myrsky on sielusi myllertänyt.
Sa iloitse, rauhan sait!
Ei riemus tornikorkeat kuohut käy.
Sa iloitse, rauhan sait!
Ei syöstä sua tuskien syvyyksiin,
ei koskaan, ei!
Laps, olet lempinyt; koskaan ei sun sielusi loimua
nyt. Sinä olithan kuin kulo kuivunut ja hetkeksi
liekkihin leimahdit. Tieltä tuhkan ja sauhun tupruavan
lens' taivahan linnut kirkuen pois. Ne palatkoot! Enää
sa et palaa, et —

Laps, olet lempinyt; koskaan et nyt rakkauden
ääntä kuulla saa! Sun sydämesi voima kuin uupunut
laps, joka koulunpenkillään ikävöi ulos vapauteen
sekä leikkiin, mut kukaan ei kutsu nyt. Niin on se kuin
vartio unhoon jäänyt: ei kutsuta, ei!
Laps, ainoas mennyt on, mennyt kera rakkaus
kaikki ja riemu sen. Hän, jota lemmit, kuin neuvonut
ois sinut siivin hän ilmoja liitelemään! Jota lemmit,
kuin ainoan turvatun paikan kyläss' antanut sulle hän
hukkuvass' ois, hän on mennyt, hän joka yksin sydän-
uksesi avata ties.
Yhtä ainoista anelen sulta, mun rakkaani: älä koskaan
vihallas minua kuormaa; Juur heikoin heikoista maan eikö ole
se ihmissydän? Miten voisi se elää viiltävän mietteen alla, että
vaivaksi toisen se oisi?
Minun rakkaani, jos sinä surmata mielit,
ei tikari tarpeen, äl' osta myrkkyä, köyttäkään!
Mun suo vain tietää, että mun poies sa tahdot
maan vihreän piiristä, valtakunnista elon.
Ja kohta ma hautaani vaivun!
Elon elämän annoit mulle. Rakkauden soit.
Nyt pois otat lahjasi! Oi, hyvin tiedän sen.
Mut vihaksi sit' elä muuta!
Rakas sentään on elo mulle. Se muista, oi!
Mut tiedän: ma kuolen vihan kuorman alle!

Kymmenes luku
NUORI KREIVITÄR
Nuori kreivitär makaa kello kymmeneen aamulla ja haluaa, että
joka päivä on tuoretta leipää aamiaispöydässä. Nuori kreivitär
harrastaa kehäompelua ja runoutta. Hän ei piittaa kutomisesta eikä
ruoanlaitosta. Nuori kreivitär on hemmoteltu.
Mutta nuori kreivitär on iloinen ja antaa ilonsa paistaa kaikkeen ja
kaikkialle. Annetaan niin mielellään anteeksi hänen pitkä aamu-
unensa ja hänen tuore leipänsä, sillä hän on tavattoman aulis
köyhille ja ystävällinen kaikille.
Nuoren kreivittären isä on ruotsalainen aatelismies, joka on asunut
koko ikänsä Italiassa, missä häntä on pidättänyt kaunis maa ja
kauniin maan kaunis tytär. Kun kreivi Henrik Dohna matkusti Italiaan,
hän joutui tämän ylimyksen kotiin, tutustui hänen tyttäriinsä, nai
yhden heistä ja toi hänet mukanaan Ruotsiin.
Kreivitär, joka oli aina osannut ruotsia ja oli kasvatettu
rakastamaan kaikkea ruotsalaista, viihtyy hyvin karhujen maan
perukassa. Hän pyörähtelee niin iloisesti siinä pitkässä huvien
katrillissa, joka viuhuu ympäri Lövenin pitkän järven rantoja, että
luulisi hänen aina eläneen täällä. Vähän hän tosiaan ymmärtää, mitä
on kreivittären arvo. Ei komeuden tavoittelua, ei jäykkyyttä, ei
alentuvaa arvokkuutta ole tässä nuoressa, iloisessa olennossa.
Vanhat herrat pitivät enimmän nuoresta kreivittärestä. Oli
merkillistä, millainen menestys hänellä oli näiden piirissä. Kun he
näkivät hänet tanssiaisissa, voi olla varma, että he kaikki, niin hyvin

Munkerudin laamanni ja Bron rovasti kuin Melchior Sinclairekin ja
Bergan kapteeni, selittivät rouvilleen mitä luottamuksellisimmin, että
jos he olisivat tavanneet nuoren kreivittären neljäkymmentä tai
kolmekymmentä vuotta sitten…
Niin, silloin hän tosin ei ollut vielä syntynyt, sanoivat vanhat
rouvat.
Ja seuraavalla kerralla nuoren kreivittären tavatessaan
kiusottelevat rouvat häntä, että hän ryöstää heiltä vanhain herrain
sydämet.
Vanhat rouvat katselevat häntä ikäänkuin huolissaan. He muistavat
niin hyvin kreivitär Märtan. Yhtä iloinen ja hyvä ja rakastettu oli
hänkin ollut tullessaan ensi kerran Bergaan, Ja hänestä oli tullut vain
turhamainen ja huvinhaluinen koketti, joka ei nyt voi ajatella mitään
muuta kuin huvejansa. Olisipa hänellä mies, joka panisi hänet
työhön! sanovat vanhat rouvat. Jospa hän osaisi kutoa kangasta!
Sillä kankaankutominen lohduttaa kaikissa suruissa, se nielee kaikki
harrastukset, se on ollut monen naisen pelastus.
Nuori kreivitär tahtoo mielellään tulla hyväksi perheenäidiksi. Hän
ei tiedä mitään parempaa kuin onnellisen vaimon elämä hyvässä
kodissa, ja hän tulee usein suurissa pidoissa istumaan vanhusten
joukkoon.
Henrik tahtoisi niin kovin, että minusta tulisi yhtä hyvä emäntä,
sanoo hän, kuin hänen äitinsä oli. Opettakaa minua kutomaan
kankaita.
Silloin huokaavat vanhat rouvat kaksinaisesti: ensiksi Henrik-
kreivin tähden, joka voi uskoa, että hänen äitinsä oli kunnon emäntä,

ja toiseksi suurien vaikeuksien vuoksi, joihin joutuisi, jos ryhtyisi
opettamaan tälle nuorelle, tietämättömälle olennolle niin mutkallisia
salaisuuksia. Tarvitsi virkkaa hänelle vain pasmasta ja tutkaimesta,
niidestä ja haasta, yksiniitisestä ja kaksivartisesta, niin hänen päänsä
jo meni pyörälle; mitä sitten ridantoimikkaasta ja hanhensilmästä ja
kilpikankaasta!
Kukaan, joka vain nuoren kreivittären näkee, ei voi olla
ihmettelemättä, miksi hän on mennyt tuhmalle Henrik-kreiville.
Voi sitä raukkaa, joka on niin tuhma! Sellaista on aina surku. Ja
suurin surku on tyhmyriä, joka elää Vermlannissa.
Jo on monta juttua Henrik-kreivin tuhmuudesta, ja hän on vasta
muutamia vuosia kolmannellakymmentä. Kerrotaan esimerkiksi,
miten hän muutama vuosi sitten huvitti Anna Stjärnhökiä
rekiretkellä.
Sinä olet kaunis, Anna, hän sanoi.
Oh, loruja, Henrik.
Olet kaunein koko Vermlannissa.
Enpäs olekaan.
Kaunein tällä rekiretkellä sinä ainakin olet.
Ah, Henrik, en minä ole sitäkään.
Niin, mutta kyllä sinä ainakin tässä reessä olet kaunein. Sitä et
ainakaan voi kieltää.
Ei, sitä ei Anna voinut.

Sillä Henrik-kreivi ei ole kaunis, ei. Hän on yhtä ruma kuin
tuhmakin. Sanotaan, että se pää, joka on hänen ohuen kaulansa
päässä, on kulkenut perintönä suvussa jo pari sataa vuotta.
Sentähden ovat viimeisen perillisen aivot niin lopen kuluneet.
Näkeehän sen, ettei hänellä ole omaa päätä, sanotaan. Hän on
lainannut pään isältään. Hänhän ei uskalla sitä taivuttaa. Pelkää että
se putoaa. — Hänellähän on jo keltainen iho ja ryppyinen otsa.
Päätä on varmaankin käyttänyt sekä isä että isoisä. Miksi tukka olisi
muutoin niin ohut ja huulet niin verettömät ja leuka niin kiverä?
Aina hänen ympärillään oli koiranleukoja, jotka puijasivat hänet
lausumaan tuhmuuksia ja sitten pistivät ne mieleensä, levittivät ja
parantelivat niitä.
Onnekseen hän ei huomaa mitään. Hänen käytöksensä on
juhlallista ja arvokasta. Voiko hän aavistaa, etteivät toisetkin olisi
sellaisia? Arvokkuus on pinttynyt hänen nahkaansa: hän liikkuu aina
samalla tavoin, tikkusuorana, eikä koskaan käännä päätään koko
ruumiin samalla kääntymättä.
Hän oli tullut kyläilylle Munkerudin laamannin luo joku vuosi sitten.
Ratsain hän oli tullut, hänellä oli ollut korkea hattu, keltaiset housut
ja välkkyvät saappaat ja niin oli istunut jäykkänä ja uljaana
satulassa. Tulo onnistui kyllä. Mutta kun hänen piti lähteä talosta,
sattui, että koivukäytävässä oksa pyyhkäisi häneltä hatun päästä.
Hän laskeutui maahan, otti hattunsa ja ratsasti taas saman oksan
alitse. Taas pyyhkäisi oksa hatun pois. Temppu uudistui neljä kertaa.
Laamanni tuli viimein hänen luokseen ja sanoi: Eikö veli ensi
kerralla voisi ratsastaa oksan ohitse?
Viidennellä kerralla hän pääsikin onnellisesti oksan ohi.

Mutta asia oli kuitenkin niin, että nuori kreivitär hänen ukon-
päästään huolimatta piti hänestä. Eihän neito tiennyt nähdessään
hänet Roomassa, että häntä kotimaassa ympäröi moinen tuhmuuden
marttyyrikunnia. Siellä oli kreivissä ollut ikäänkuin
nuoruudenloistetta, ja he olivat yhtyneet toisiinsa kovin romanttisissa
oloissa. Tarvitsipa vain kuulla kreivittären kertovan, miten Henrik-
kreivin oli täytynyt hänet ryöstää. Munkit ja kardinaalit olivat
vihastuneet kauheasti siitä, että kreivitär aikoi luopua äitinsä
uskosta, johon oli ennen kuulunut, ja kääntyä protestantiksi. Koko
roskaväki oli kuohuksissaan. Tytön isän palatsia piiritettiin. Roistot
ajelivat Henrikiä. Äiti ja sisar rukoilivat häntä luopumaan
avioaikeistaan. Mutta hänen isänsä raivostui siitä, että italialainen
roskajoukko muka aikoi estää häntä antamasta tytärtään kenelle hän
tahtoi. Isä käski Henrik-kreivin ryöstämään tyttären. Ja niinpä he,
kun oli mahdoton vihkiä heitä kotona väen huomaamatta, niinpä hän
ja Henrik hiipivät takakatuja ja kaikkia mahdollisia pimeitä teitä
Ruotsin konsulinvirastoon. Ja kun kreivitär siellä oli luopunut
katolilaisuudestaan ja tullut protestantiksi, vihittiin heidät tuossa
tuokiossa ja lähetettiin kiitävissä matkavaunuissa pohjoiseen. Siinä
ei jouduttu kuuluttamaan eikä muuta, näettekös. Se oli aivan
mahdotonta, on nuoren kreivittären tapana sanoa. Ja olihan
tietenkin synkkää, että meidät vihittiin konsulinvirastossa eikä siellä
kauniissa kirkossa, mutta muuten olisi Henrik jäänyt ilman minua.
Siellä ovat kaikki niin kiivaita, sekä isä että äiti, ja kardinaalit ja
munkit, kaikki ovat kiivaita. Siksi sen täytyi käydä niin kovin
salaisesti, ja jos väki olisi nähnyt meidän hiipivän kadulle kotoamme,
se olisi varmaan tappanut meidät kummatkin — pelastaakseen
minun sieluni. Henrik oli tietysti jo kadotettu.
Mutta nuori kreivitär pitää miehestään yhä vielä, heidän tultuaan
Borgiin ja rauhallisempaan elämään. Hän rakastaa hänessä vanhan

nimen loistoa ja mainehikkaita esi-isiä. Hänestä on suloista nähdä
oman läsnäolonsa pehmittävän kreivin synnynnäistä jäykkyyttä ja
kuulla, miten kreivin ääni lämpenee heidän puhellessaan keskenään.
Ja sitä paitsi kreivi pitää hänestä ja hemmottelee häntä, ja onhan
hän nyt kerran naimisissa hänen kanssaan. Nuori kreivitär ei juuri voi
ajatellakaan, ettei nainen rakastaisi miestään.
Jollakin tavalla kreivi vastaa myöskin hänen miehuuden
ihannettaan. Hän on rehellinen ja totuutta rakastava. Hän ei ole
koskaan syönyt sanaansa. Kreivittärestä hän on todellinen ylimys.
* * * * *
Maaliskuun kahdeksantena viettää nimismies Scharling
syntymäpäiviään, ja silloin nousee väkeä Brobyn ylämäkeä. Ihmisiä
idästä ja lännestä, tuttuja ja tuntemattomia, kutsuttuja ja
kuokkavieraita tulee tavallisesti silloin nimismiehen taloon. Kaikki
ovat tervetulleita. Kaikille on kylliksi ruokaa ja juomaa, ja
tanssisalissa on tilaa seitsemän pitäjän tanssihaluisille.
Nuori kreivitär tulee myös, niinkuin hän tulee kaikkialle, missä
tanssia ja huvia voi odottaa.
Mutta nuori kreivitär ei tullessaan ole iloinen. Hän ikäänkuin
aavistaa, että nyt on hänen vuoronsa tempautua seikkailujen
hurjaan ajojahtiin.
Matkalla hän on katsellut laskeutuvaa aurinkoa. Se vaipui alas
pilvettömältä taivaalta eikä jättänyt mennessään kultareunoja
hattaroihin. Harmaankelmeä hämärä, kylmien myrskypuuskien
halkoma, himmensi maan.

Nuori kreivitär katseli, miten päivä ja yö keskenään taistelivat ja
miten pelko valtasi kaiken elollisen tuota mahtavain taisteloa
katsellessaan. Hevoset leiskoivat nopeammin viimeistä kuormaansa
joutuakseen pian suojaan. Hirrenhakkaajat kiiruhtivat kotiin
metsästä, piiat karjakartanoista. Pedot ulvoivat metsänrinnassa.
Päivä, ihmisten lemmikki, joutui tappiolle.
Valo sammui, värit vaalenivat. Kylmää ja rumaa oli kaikki mitä hän
näki. Mitä hän oli toivonut, mitä rakastanut, mitä tehnyt, kaikki näytti
hänestä verhoutuvan harmaaseen hämärän huntuun. Se oli
väsymyksen, tappion, voimattomuuden hetki hänelle niinkuin koko
luonnolle.
Hän ajatteli, että hänen oma sydämensä, joka nyt räiskyvässä
riemussaan verhosi elämän purppuralla ja kullalla, ehkä kerran
kadottaa voimansa eikä jaksa enää valaista hänen maailmaansa.
Oi, avuttomuus, oman sydämeni voimattomuus! sanoi hän
itsekseen. Tukahduttavan harmaan hämärän jumalatar, kerran olet
sinä sieluni valtiatar. Silloin näen elämän rumana ja harmaana,
jollaista se kenties onkin, silloin tukkani valkenee, selkäni koukistuu,
aivoni lamautuvat.
Samassa kiepahti reki nimismiehen pihaan, ja kun nuori kreivitär
juuri nosti silmänsä, sattui hänen katseensa sivurakennuksen
rautaristikkoiseen ikkunaan, ja hän näki sen takana tuimat
ihmiskasvot.
Ne olivat Ekebyn majurinrouvan kasvot, ja nuori kreivitär tunsi,
että nyt oli hänen iltainen ilonsa turmeltu.

Käyhän hyvin laatuun olla iloinen, kun ei surua näe, vaan kuulee
vain siitä puhuttavan kuin muukalaisesta vieraalla maalla. Pahempi
on säilyttää sydämensä iloa seistessään silmätysten yömustan,
tuimasti tuijottavan tuskan kanssa.
Kreivitär tietää kyllä, että nimismies Scharling on pannut
majurinrouvan putkaan ja että tämä joutuu tutkittavaksi niiden
väkivaltaisuuksien tähden, jotka hän sai aikaan Ekebyssä sinä yönä,
jolloin ne suuret tanssiaiset olivat. Mutta hän ei juuri ajatellut, että
majurinrouvaa pidettäisiin nimismiehen talossa ja niin lähellä
tanssisalia, että sieltä saattoi silmäillä hänen huoneeseensa; niin
lähellä, että hän varmaan kuuli tanssimusiikin ja iloisen hälinän. Ja
nyt ryöstää majurinrouvan ajatteleminen kreivittäreltä kaiken ilon.
Nuori kreivitär tanssii kyllä sekä valssit että katrillit. Hän liehuu
kyllä sekä menuetissa että angleesissa, mutta jokaisen tanssin
loputtua hänen täytyy hiipiä ikkunaan katselemaan pihan toiseen
laitaan, sivurakennukseen. Kynttilä palaa majurinrouvan ikkunassa,
ja kreivitär näkee tämän käyskentelevän edestakaisin huoneessa.
Hän ei näytä lepäävän hetkeäkään, vaan kävelee lakkaamatta.
Kreivitär ei nyt iloitse tanssista. Hän ajattelee vain, että
majurinrouva kävelee edestakaisin vankilassaan kuin häkkiin pantu
villipeto. Hän ihmettelee, miten kaikki muut voivat tanssia. Varmaan
siellä on monta, joiden mieltä ahdistaa yhtä paljon kuin hänenkin,
kun tietävät majurinrouvan olevan niin lähellä, ja kumminkaan ei
kukaan ilmaise ajatustaan. Vermlannissa asuu suvaitsevaista kansaa.
Mutta aina kun hän katsoo ulos, liikkuvat hänen jalkansa tanssissa
raskaammin, ja nauru on jähmettyä hänen kurkkuunsa.

Nimismiehen rouva tarkkaa häntä, kun hän ulos nähdäkseen
pyyhkii huurua ikkunanruudusta, ja tulee hänen luokseen.
Sellaista kurjuutta! Voi miten kurjaa tämä on! kuiskaa hän
kreivittärelle.
Minusta on melkein mahdotonta tanssia tänä iltana, kuiskaa
kreivitär puolestaan.
Tanssiaiset eivät nyt olekaan minun tahdostani, kun hän on tuolla
vankina, vastaa rouva Scharling. Hän on ollut Karlstadissa koko
ajan vangitsemisesta lähtien. Nyt tulee pian tutkinto, ja senvuoksi
hänet tuotiin tänään tänne. Emme voineet panna häntä käräjätalon
viheliäiseen putkaan, vaan sijoitimme hänet sivurakennuksen
kutomahuoneeseen. Hän olisi saanut olla vierashuoneessani,
kreivitär, jollei tämä vierasjoukko olisi tullut juuri tänään. Kreivitär
tuskin tuntee häntä, mutta hän on ollut kuin meidän kaikkien äiti ja
kuningatar. Mitä hän ajatteleekaan meistä, kun me täällä tanssimme
hänen itsensä ollessa niin suuressa hädässä. Hyvä on, että useimmat
eivät tiedä hänen olevan siellä.
Häntä ei olisi pitänyt vangitakaan, sanoo nuori kreivitär
ankarasti.
Niin, se on ihan totta, kreivitär, mutta ei ollut muuta keinoa, jos
mieli välttää pahempia onnettomuuksia. Ei ollut ketään, joka
paheksui sitä, että hän pani omat olkiaumansa tuleen ja tahtoi
karkottaa kavaljeerit, mutta majurihan samoili ajelemassa häntä.
Jumala tietää, mitä majuri olisi hänelle tehnyt, jos häntä ei olisi
pantu kiinni. Scharling on saanut paljon ikävyyksiä siitä, että vangitsi
majurinrouvan. Yksinpä Karlstadissakin oltiin häneen tyytymättömiä,

kun hän ei painanut Ekebyn tapahtumia villaisella; mutta hän teki
minkä parhaaksi näki.
Mutta nyt hänet tuomitaan rangaistukseen, sanoi kreivitär.
Oh ei, kreivitär, ei häntä tuomita. Ekebyn majurinrouva pääsee
kyllä vapaaksi, mutta jo tämäkin, mitä hän on saanut näinä päivinä
kestää, on hänelle liikaa. Hän tulee varmaan hassuksi. Arvaattehan,
kuinka niin ylpeä rouva voisi alistua kohdeltavaksi kuin rikollinen!
Minusta olisi ollut parasta, että hän olisi saanut olla vapaana. Hän
olisi ehkä itse välttänyt vaaran.
Päästäkää hänet vapaaksi, sanoo kreivitär.
Sen kyllä voi tehdä kuka muu tahansa paitsi nimismies ja hänen
vaimonsa, kuiskaa rouva Scharling. Täytyyhän meidän vartioida
häntä, meidän. Ja erityisesti tänä yönä, jolloin täällä on niin monta
hänen ystäväänsä, istuu kaksi miestä vartioimassa hänen oveaan,
joka on lukittu ja salvattu niin, ettei kukaan voi päästä hänen
luokseen. Mutta jos joku päästäisi hänet sieltä, kreivitär, niin kyllä
sekä Scharling että minä olisimme iloisia.
Enkö minä saa mennä hänen luokseen? kysyy nuori kreivitär.
Rouva Scharling ottaa innokkaasti häntä ranteesta ja vie hänet
salista. Eteisessä he pistävät hartioilleen saalit, ja niin he kiiruhtavat
pihan poikki.
Ei ole varmaa, puhuuko hän edes meille, sanoo nimismiehen
rouva.
Mutta saahan hän kumminkin nähdä, ettemme ole unohtaneet
häntä.

He tulevat rakennuksen ensimmäiseen huoneeseen, jossa nuo
kaksi miestä istuvat vartioimassa salvattua ovea, ja pääsevät esteittä
majurinrouvan luo. Tämä on suuressa huoneessa, joka on täynnä
kangaspuita ja muita työkaluja. Huone on oikeastaan kutomakamari,
mutta sen ikkunassa on rautaristikko ja ovessa vankat lukot, niin että
sitä hätätilassa voi käyttää putkana.
Majurinrouva jatkoi kävelyään tuskin osoittamatta heille huomiota.
Hän on pitkällä matkalla, ollut jo monta päivää. Hän ei muista
muuta kuin että hän astuu vain niitä kahtakymmentä peninkulmaa
äitinsä luo, joka elelee kaukana Älfdalin metsissä häntä odottaen.
Hänellä ei ole aikaa levätä. Tavaton kiire hänellä on. Hänen äitinsä
on yli yhdeksänkymmenen vanha. Hän kuolee varmaan pian.
Hän on mitannut lattian pituuden kyynäräkepillä, ja nyt hän laskee
kerrat yhteen, kyynärät syliksi ja sylet puolipeninkulmiksi ja
kokopeninkulmiksi.
Raskaalta ja pitkältä hänestä tuntuu tämä kulku eikä hän
kumminkaan uskalla levätä. Hän kahlaa syvissä kinoksissa. Hän
kuulee kulkiessaan metsien ikuisen huminan yllään. Hän huoahtaa
suomalaisen pirtissä ja hiilenpolttajan risumajassa. Joskus, kun ei ole
yhtään ihmistä peninkulmain taipaleella, hänen täytyy katkoa oksia
alleen ja levähtää kaatuneen kuusen juurikon suojassa.
Ja vihdoinkin hän on päässyt perille, taivaltanut kaksikymmentä
peninkulmaa, metsä aukenee, ja punaiset asuinrakennukset
kyyhöttävät lumisen pihan ympärillä. Klara-joki virtaa ohi vaahtona
pärskyen, useina pieninä koskina, ja tästä tutusta pauhusta hän
kuulee, että on kotona. Ja hänen äitinsä, nähdessään hänen tulevan
kerjäläisenä, kuten oli tahtonutkin, tulee häntä vastaan.

Kun majurinrouva on päässyt niin pitkälle, hän nostaa aina
päätään, katsoo ympärilleen, näkee suljetun oven ja muistaa missä
on.
Silloin hän ajattelee, että hulluksiko hän on tulemaisillaan, ja
istuutuu miettimään muuta ja lepäämään. Mutta hetken perästä hän
on taas matkalla, hän laskee kyynäristä puoli- ja kokopeninkulmia,
lepää hetkisen suomalaisten tuvissa eikä makaa yöllä eikä päivällä
ennen kuin on kulkenut nuo kaksikymmentä peninkulmaa.
Koko vankeutensa aikana hän ei ole juuri ollenkaan nukkunut.
Ja naiset, jotka tulivat hänen luokseen, katsovat häntä tuskaisina.
Nuori kreivitär muistaa sittemmin hänet aina sellaisena kuin hän
oli siellä kävellessään. Hän näkee hänet usein unissaan ja herää
näystään kyyneleisin silmin ja valitus huulillaan.
Vanhus on surkeasti mennyt alaspäin, tukka näyttää ohuelta, ja
irtonaisia hiustupsuja siirottaa hänen vähäisestä palmikostaan.
Kasvot ovat raukeat ja painuneet, rääsyiset vaatteet riippuvat
rumasti hänen yllään. Mutta kaikesta huolimatta hänessä on niin
paljon ylevää, käskyillään kaikki alistavaa valtiatarta, ettei hän herätä
ainoastaan sääliä, vaan myöskin kunnioitusta.
Mutta kaikkein selvimmin muisti kreivitär hänen silmänsä, jotka
olivat kuoppiinsa painuneet ja omiin mietteihinsä kääntyneet ja joista
ei vielä ollut kaikki järjen valo sammunut, vaikka oli melkein
sammumaisillaan, ja joissa vaani syvällä sellainen hurjuuden kipinä
että kauhistutti ja pelotti, että vanhus tuossa tuokiossa hyökkää
kimppuun, puree hampaillaan, kynsii sormillaan.

He ovat nyt seisoneet siellä kotvan aikaa, kun majurinrouva
yht'äkkiä pysähtyy nuoren kreivittären eteen ja katsoo häntä hyvin
ankarasti. Kreivitär peräytyy askelen ja tarttuu rouva Scharlingia
käsivarteen.
Majurinrouvan kasvot elostuvat yht'äkkiä ja selkenevät, hänen
silmänsä katsovat maailmaan täysin tajuisesti.
Oh, ei, ei, hän sanoo hymyillen, niin hullusti ei toki ole laita,
rakas nuori rouva.
Hän pyytää nyt heitä istumaan ja istuu itsekin. Hänen kasvoilleen
tulee entinen komea ilme, samanlainen kuin Ekebyn suurissa
kemuissa ja kuninkaan tanssiaisissa Karlstadin maaherrankartanossa.
Toiset unohtavat rääsyt ja vankilan ja näkevät ainoastaan Vermlannin
ylpeimmän ja rikkaimman naisen.
Rakas kreivittäreni, hän sanoo. Mikä saattaa teidät
keskeyttämään tanssinne ja tulemaan minunlaiseni yksinäisen
mummon luo? Taidatte olla varsin hyvä.
Elisabet-kreivitär ei voi vastata. Liikutus tukkii hänen äänensä.
Rouva Scharling vastaa hänen puolestaan, ettei kreivitär ole voinut
tanssia ajatellessaan majurinrouvaa.
Rakas rouva Scharling, vastaa majurinrouva, olenkin minä jo
niin surkealla tolalla, että häiritsen nuorten ilonpitoa? Ette saa itkeä
minun vuokseni, nuori kreivitär, hän jatkoi. Olen vanha, ilkeä
nainen, joka ansaitsen kohtaloni. Teidän mielestänne ei kai tee oikein
se, joka lyö äitiään?
Ei, mutta…

Majurinrouva keskeyttää hänet ja pyyhkäisee hänen otsaltaan
kiharaisen, vaalean tukan.
Lapsi, lapsi, hän sanoo, miten te saatoitte ottaa tuhman Henrik
Dohnan?
Mutta minä rakastan häntä.
Näen miten on, näen miten on, sanoo majurinrouva. Kiltti lapsi,
eikä muuta: itkee surullisten ja nauraa iloisten kanssa. Ja pakostakin
sanoo 'jaa' ensimmäiselle, joka sanoo: 'Minä rakastan sinua'! Niin,
niinpä niin. Menkää nyt takaisin ja tanssikaa, rakas nuori kreivitär.
Tanssikaa te vain ja olkaa iloinen. Teissä ei ole pahuutta.
Mutta minä tahtoisin tehdä jotain teidän hyväksenne,
majurinrouva.
Lapseni, sanoo majurinrouva juhlallisesti, Ekebyssä asui vanha
nainen, jolla oli taivaan tuuletkin vankinaan. Nyt hän on itse vankina,
ja tuulet irrallaan. Onko ihme, että myrsky maassa mylvii?
Minä, joka olen vanha, näin sen jo ennen. Tunsin sen. Tiedän,
että Jumalan jyrisevä myrsky tulee päällemme. Milloin se suhisee
suurissa valtakunnissa, milloin viuhuu pienten syrjäisten
yhteiskuntain kimpussa. Jumalan myrsky ei unohda ketään. Se
kohtaa niin suuria kuin pieniä. On komeaa, tuo Jumalan myrskyn
tulo.
Jumalan myrsky, sinä siunattu Herran ilma, puhalla maassa!
Äänet ilmassa, äänet maassa, kaikukaa ja kauhistakaa! Tehkää
Jumalan myrsky jyriseväksi! Tehkää Jumalan myrsky hirvittäväksi!
Kiitäkööt myrskynpuuskat maassa, syöksykööt horjuviin seiniin,

murtakoot lukot, jotka ovat ruostuneet, ja huoneet, jotka kallistuvat
kaatuakseen.
Kauhu on täyttävä maan. Pienet linnunpesät putoavat oksainsa
varasta puista maahan. Haukanpesä honganlatvasta putoaa suurella
pauhulla alas, ja huuhkajankin pesään, vuorenrotkoon, sähisee tuuli
lohikäärmeenkielin.
Me luulimme, että meillä on kaikki hyvin täällä; mutta niin ei ollut.
Jumalan myrskyä tarvitaan. Minä ymmärrän sen enkä valita.
Tahtoisin vain mennä äitini luo.
Hän painuu yht'äkkiä kyyryyn.
Mene nyt, nuori nainen, hän sanoo. Minulla ei ole enää aikaa.
Minun täytyy lähteä. Menkää nyt ja varokaa niitä, jotka ratsastavat
myrskypilvillä!
Ja sitten hän alkaa taas vaelluksensa. Piirteet höltyvät, katse
kääntyy sisäänpäin. Kreivittären ja rouva Scharlingin täytyy jättää
hänet.
Heti kun he ehtivät takaisin tanssijain joukkoon, menee nuori
kreivitär
Gösta Berlingin luo.
Tuon teille, herra Berling, terveisiä majurinrouvalta, hän sanoo.
Hän odottaa, että herra Berling laskee hänet vankeudesta.
Saapa sitten odottaa, kreivitär.
Oi, auttakaa häntä, herra Berling!

Gösta silmäilee eteensä synkästi. Ei, sanoo hän, miksi minä
auttaisin häntä? Mistä minä olen hänelle kiitollisuuden velassa?
Kaikki, mitä hän on tehnyt hyväkseni, on ollut turmiokseni.
Mutta, herra Berling…
Jos häntä ei olisi ollut, sanoo Gösta kiivaasti, makaisin nyt tuolla
ikuisten metsien helmassa. Olenko minä velvollinen uskaltamaan
henkeni hänen tähtensä siksi, että hän teki minusta Ekebyn
kavaljeerin? Luuleeko kreivitär, että siitä virasta suurta kunniaa
koituu?
Nuori kreivitär kääntyy hänestä vastaamatta. Hän on vihainen.
Hän palaa paikalleen, katkeroituneena kavaljeereihin. Tänne he
ovat tulleet walt-torvineen ja viuluineen ja aikovat antaa käyrien
hangata kieliä, kunnes jouhet kuluvat poikki, ajattelematta, että ilon
soitto kuuluu pihan yli vangin viheliäiseen huoneeseen. Tänne he
tulevat tanssimaan anturansa tomuksi eivätkä ajattele, että heidän
vanha hyväntekijänsä voi nähdä heidän varjojensa leijailevan ohi
huuruisten ruutujen! Ah, miten maailma tuli harmaaksi ja rumaksi!
Ah, minkä varjon hätä ja kovuus heitti nuoren kreivittären sieluun!
Hetken päästä tulee Gösta pyytämään häntä tanssiin. Kreivitär
kieltäytyy jyrkästi.
Ettekö tahdo tanssia minun kanssani, kreivitär? kysyy Gösta ja
tulee hyvin punaiseksi.
En teidän enkä muidenkaan Ekebyn kavaljeerien, hän sanoo.
Emme siis ole sen kunnian arvoisia?

Se ei ole mikään kunnia, herra Berling. Mutta minua ei huvita
tanssia sellaisten kanssa, jotka unohtavat kaikki kiitollisuuden
käskyt.
Gösta on jo pyörähtänyt kantapäillään. Tämän kohtauksen moni
kuulee ja näkee. Kaikki myöntävät kreivittären olevan oikeassa.
Kavaljeerien kiittämättömyys ja sydämettömyys majurinrouvaa
kohtaan on herättänyt yleistä suuttumusta.
Mutta niinä päivinä on Gösta Berling vaarallisempi kuin metsän
peto. Aina siitä lähtien kun hän tuli kotiin karhunajosta ja näki
Mariannen menneen, on hänen sydämensä ollut kuin arka haava.
Häntä haluttaisi tehdä jollekin veristä vääryyttä ja levittää laajalti
surua ja tuskaa.
Jos kreivitär niin tahtoo, hän sanoo itsekseen, tapahtukoon niin
kuin hän tahtoo. Mutta älköön kreivitär yrittäkökään säästää omaa
nahkaansa. Nuorta kreivitärtä miellyttävät naisenryöstöt. Hän saa
nyt toteuttaa mielitekonsa. Gösta Berling on perin halukas pieneen
seikkailuun. Kahdeksan päivää hän on kantanut surua naisen vuoksi.
Se jo riittää. Hän kutsuu puheilleen Beerencreutzin, everstin, ja
Kristian Berghin, vahvan kapteenin, ja hitaan Kristoffer-serkun, joka
ei ikinä siekaile heittäytyä hurjapäiseen seikkailuun, ja neuvottelee
heidän kanssaan, miten kavaljeerirakennuksen loukattu kunnia on
kostettava.
Sitten loppuvat pidot. Pitkä jono rekiä ajetaan pihalle. Herrat
vetävät turkit yllensä. Naiset etsivät vaatteitaan pukuhuoneen
toivottomasta sekamelskasta.
Nuori kreivitär pitää kiirettä päästäkseen jo näistä inhottavista
tanssiaisista. Hän joutuu naisista ensimmäisenä valmiiksi. Hän seisoo

hymyillen keskellä naistenhuoneen lattiaa ja katsoo muiden
lähtötouhua, kun yht'äkkiä ovi aukeaa ja Gösta Berling seisoo
kynnyksellä.
Miehillä ei tietenkään ole oikeutta tunkeutua tähän huoneeseen.
Vanhat rouvat seisovat siellä vähine hiuksineen, pantuaan koristavat
myssynsä pois. Ja nuoret ovat kääntäneet hameensa helmat turkin
alle, etteivät jäykät volangit rutistuisi matkalla.
Mutta hillitsevistä huudoista välittämättä hyökkää Gösta Berling
kreivittären luo ja tempaa hänet.
Ja hän nostaa hänet käsivarsilleen ja ryntää kamarista eteiseen ja
sieltä portaille.
Eivät voi ällistyneiden naisten huudot pysähdyttää häntä. Kun he
juoksevat perästä, he näkevät vain, miten hän viskautuu rekeensä
kreivitär sylissä.
He kuulevat kuskin piiskanläiskeen ja näkevät hevosen
porhaltavan laukkaan. He tuntevat kuskin — se on Beerencreutz. He
tuntevat hevosen — se on Don Juan.
Ja kreivittären kohtalosta kovin huolissaan he huutavat herroja.
Ja nämä eivät hukkaa aikaa moniin kysymyksiin, vaan syöksyvät
rekiinsä.
Ja kreivi etunenässä he ajavat takaa naisenryöstäjiä.
Mutta Gösta Berling kellettää reessä pitäen kiinni nuorta
kreivitärtä. Surut kaikki hän on unohtanut, ja seikkailun
juovuttavasta riemusta huimana hän laulaa täyttä kurkkua laulua
rakkaudesta ja ruusuista.

Aivan rintaansa vasten hän puristaa kreivitärtä; mutta kreivitär ei
yritäkään paeta. Hänen kasvonsa lepäävät valkeina ja jäykistyneinä
Göstan rintaa vasten.
Ah, mitä tulee miehen tehdä, kun häntä liki on naisen kalpeat,
avuttomat kasvot, kun mies näkee vaalean tukan, joka tavallisesti
varjostaa valkeata, loistavaa otsaa, mutta on nyt syrjään häilähtänyt,
ja kun silmäluomet ovat raskaasti sulkeutuneet peittäen harmaiden
silmien veitikkakimelteen.
Mitä tulee miehen tehdä, kun punaiset huulet vaalenevat hänen
silmiensä edessä?
Suudella, tietysti suudella noita vaalenevia huulia, noita suljettuja
silmiä, tuota valkoista otsaa.
Mutta silloin herää nuori nainen. Hän tahtoo kimmahtaa pois. Hän
on kuin viritetty jousi. Ja Göstan on kamppailtava kaikin voimin, ettei
toinen pääsisi heittäytymään reestä, kunnes hän on saanut
kreivittären kukistettuna ja vapisevana vaipumaan reen toiseen
nurkkaan.
Katso, sanoo Gösta silloin aivan tyynesti Beerencreutzille.
Kreivitär on kolmas, jota Don Juan ja minä viemme tänä talvena.
Mutta ne toiset riippuivat suudellen kaulassani, kun tämä taas ei
tahdo antaa minun suudella eikä tanssiakaan kanssaan. Tuletkos
hullua hurskaammaksi noista naisista, Beerencreutz?
Mutta Göstan ajaessa pihasta ja naisten kirkuessa ja huudellessa
hädissään ja touhuissaan joutuivat majurinrouvaa vartioivat miehet
ihmeisiinsä.

Mikäs nyt on? he ajattelivat. Mistä moinen kirkuminen?
Samassa lyödään ovi auki, ja ääni huutaa heille:
Hän on poissa! Nyt se vie häntä.
He ulos, juosten kuin hullut, ajattelematta, majurinrouvako vai
kuka se oli poissa. Onnikin suosi heitä, niin että he pääsivät
erääseen ohikiitävään rekeen. Ja he ajoivat pitkän aikaa, ennen kuin
saivat tietää, ketä ajettiin.
Mutta Bergh ja Kristoffer-serkku menivät aivan rauhassa vankilan
ovelle, mursivat lukon ja avasivat oven.
Majurinrouva on vapaa, he sanoivat.
Hän tuli ulos. He seisoivat suorina kuin puikot oven kahden
puolen, katsomatta häneen.
Majurinrouvaa odottaa hevonen ja reki pihalla.
Hän meni silloin ulos, istuutui rekeen ja ajoi tiehensä. Kukaan ei
ajanut häntä takaa. Kukaan ei edes tiennyt minne hän meni.
Brobyn mäeltä rientää Don Juan Lövenin jääpeittoiselle pinnalle.
Uljas juoksija kiitää kuin lentämällä.
Raikas jääkylmä ilma viuhuu ajajien poskilla. Kulkuset helkkyvät.
Tähdet ja kuu välkkyvät. Lumi lepää sinivalkoisena ja loistaa omaa
valoaan.
Gösta tuntee runollisten ajatusten heräävän mielessään.
Beerencreutz, hän sanoo, katso, tämä on elämää. Niinkuin Don
Juan kiitää pois tätä nuorta naista vieden, niin kiitää myös aika

vieden ihmistä. Sinä olet välttämättömyys, joka ohjaat matkaa. Minä
olen himo, joka vangitsen tahdon. Ja niin temmataan hän, tuo
voimaton, yhä syvemmälle alas.
Älä lörpöttele! karjaisee Beerencreutz. Nyt ne tulevat perästä.
Ja viuhuvilla piiskanlyönneillä hän kiihoittaa Don Juania yhä
hurjempaan vauhtiin.
Siellä sudet, täällä saalis! huudahtaa Gösta. Don Juan, poikani,
kuvittele olevasi nuori hirvi. Syöksy viidakon läpi, kahlaa rämeen
poikki, hyppää tunturin selkään ja sieltä alas kirkkaaseen järveen, ui
yli, pää uljaasti koholla, ja katoa, katoa tiheän kuusikon pelastavaan
pimeään! Juokse, Don Juan, vanha naisrosvo! Juokse kuin nuori
hirvi!
Riemu täyttää hänen hurjan sydämensä vauhdin tuoksinassa.
Riemulaulua ovat hänestä vainoojien huudot. Riemu täyttää hänen
hurjan sydämensä, kun hän tuntee kreivittären ruumiin tutisevan
pelosta, kun hän kuulee hänen hampaittensa kalinan.
Yht'äkkiä irtautuu rautainen kierre, jossa hän on pitänyt
kreivitärtä.
Hän nousee pystyyn reessä ja heiluttaa lakkiaan.
Minä olen Gösta Berling, huutaa hän, kymmenen tuhannen
suutelon, kolmentoista tuhannen lemmenkirjeen herra. Eläköön,
Gösta Berling! Ottakoon kiinni ken saa!
Ja seuraavassa silmänräpäyksessä hän kuiskaa kreivittären
korvaan:

Eikö vauhti ole hyvä? Eikö tämä ole kuninkaallinen matka?
Lövenin takana Venern. Venernin takana meri, loppumattomiin aavaa
lakeutta, kirkasta, sinimustaa jäätä, ja sen tuolla puolen säteilevä
maa. Jylisevä ukkonen jäätyvän jään alla, vimman huudot
takanamme, tähdet yllämme lentävät ja kulkuset edessämme
kilisevät! Eteenpäin! Yhä eteenpäin! Haluttaako teitä, nuori, kaunis
naiseni, kokea matkan hurmaa?
Hän oli päästänyt kreivittären irti. Kreivitär sysää hänet kiivaasti
luotaan.
Seuraava hetki näkee Göstan polvillaan hänen edessään.
Olen raukka, raukka. Teidän ei olisi pitänyt, kreivitär, ärsyttää
minua. Seisoitte siellä niin ylpeänä ja hienona ettekä luullut, että
kavaljeerin koura koskaan yltää teihin. Taivas ja maa rakastaa teitä.
Teidän ei pitäisi lisätä niiden kuormaa, joita taivas ja maa
halveksivat.
Hän tempaa kreivittären kädet ja painaa ne kasvoilleen.
Jospa edes tietäisitte, sanoo hän, miltä tuntuu olla hylätty. Ei
silloin kysy mitä tekee. Ei, silloin ei kysy.
Samassa hän huomaa, ettei kreivittärellä ole käsineitä. Hän
sieppaa silloin isot rukkaset taskustaan ja pistää ne hänen käteensä.
Niin hän rauhoittuu täydellisesti. Hän laittautuu oikeaan asentoon
ja istuu nuoresta kreivittärestä niin kaukana kuin mahdollista.
Teidän ei tarvitse pelätä, kreivitär, hän sanoo. Ettekö näe mihin
tulemme? Tottahan ymmärrätte, ettemme me toki tohdi tehdä teille
mitään pahaa.

Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.
More than just a book-buying platform, we strive to be a bridge
connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.
Join us on a journey of knowledge exploration, passion nurturing, and
personal growth every day!
ebookbell.com

Imagebased Computational Modeling Of The Human Circulatory And Pulmonary Systems Methods And Applications 1st Edition Daniel R Thedens Auth

More Related Content

Similar to Imagebased Computational Modeling Of The Human Circulatory And Pulmonary Systems Methods And Applications 1st Edition Daniel R Thedens Auth (20)

Recently uploaded (20)

Imagebased Computational Modeling Of The Human Circulatory And Pulmonary Systems Methods And Applications 1st Edition Daniel R Thedens Auth