Design of Modified Bio-Inspired Algorithm for Identification and Segmentation of Pectoral Muscle in Mammogram

Design of Modified Bio-Inspired Algorithm
for Identification and Segmentation of
Pectoral Muscle in Mammogram
Presented By:
Sameer Saxena
AIIT, AUR, Jaipur
Guide: Co-Guide:
Dr. Yudhveer Singh Dr. Basant Agarwal
Associate Professor Assistant Professor
AIIT, Dept. of CS&E
AUR, Jaipur IIIT , Jaipur

Agenda
1)Introduction
2)Old Topic and preprocessing
3)Topic Modification
4)Literature
5)Problem/Challenges
6)Flow of proposed Methodology

Introduction
Cancer is a group of diseases
involving abnormal cell growth with
the potential to invade or spread to
other parts of the body.

The clinical trial indicates that early
detection and diagnosis of breast cancer
can provide early detection and
diagnosis of breast cancer can provide
patients with more flexible treatment
options and improved quality of life and
survivability.[30]

Nowadays screening
mammography is the most
adopted technique to
perform an early breast
cancer detection.
MBCD (mammographic
breast cancer diagnosis)
now supports the decision
making to differentiate
between malignant and
benign lesions by providing
additional information.[31]

 Pectoral muscle is a predominant density area
in most mammograms and may bias the
results. Its extraction can increase accuracy
and efficiency of cancer detection.

PROBLEM IDENTIFICATION
Several mammograms on which
pectoral muscle detection is difficult
using existing algorithms
(a) Pectoral muscle boundary is a
curve and becomes vertical in the
lower part (b) The intensities of
the pectoral muscle change
greatly (c) The area of the
pectoral muscle is larger than
that of the breast (d) The area of
the pectoral muscle is very small
and its boundary is nearly
vertical in the lower part (e)
Pectoral muscle mixes in the
glandular tissues in the lower
part; (f) Pectoral muscle consists
of several layers;

Challenges/Problem/Objective
In most of the mammographic images, pectoral
muscle detection still remains a challenging task.
Varying position, size, shape and texture from
image to image, textural information similar to that
of breast tissue, in most of cases.
Finding the exact pectoral muscle border in
some cases is highly difficult; especially, with
overlapping of surrounding tissues.

Testing
• Flip/Orientation the Image
• background labels/artifacts removal
• Removal of Noise
• Find boundary of pectoral muscles.

Flip/Orientation(Transformation)

Testing to find edge to remove
pectoral muscle
• SOBEL
• PREWITT
• LOG
• Roberts
• Canny

Hough Lines on Canny Edge image

Topic Modification
Design and Enhancement of Algorithm to
Identify Benign and Malignant Tumor in
Mammogram Images using
Convolutional Neural Network

Suspicious breast cancers appear
as white spots in mammograms,
indicating small clusters of micro-
calcifications.

Problem/Challanges
• The accuracy of the computer-aided systems decreases due to
some factors like density of the breast, presence of labels,
artefacts or even pectoral muscle in the mammogram image
and different types of Noises.
• Difference of perceived visual appearance between
malignant and benign lesions is unclear and consequently,
how to quantify breast lesions with discriminative features is
full of challenges. It is found that more than 70% of suggested
biopsies are with benign outcome during the diagnosis phase.
• Overfitting is a major obstacle for AI technology. Too much
knowledge will create confusion to model during training. [55]

Literature Review
(Different machine learning methodologies)
Author Dataset used Classification
Technique
Performance
Measure
j.Diz et. all
[32]
INBreast Naïve Bayes,SVM,K-
NN and DT
K-NN
classifier(77.0%
global accuracy)
j. Zzhang et. al
[33]
Private dataset Random Forest False positive
locations with 40%
Malik, Bilal et. al
[34]
QT ultrasound SVM Accuracy 90%
Muramatsu et. al.
[35]
Private dataset ANN,SVM,RF Max AUC 0.90
Saharan et. al
[36]
Breast UKM Ensemble frame
Work
Accuracy
90.8±5.0
w.Peng et al.
[37]
MIAS ANN Accuracy 96%

Comparison using different methodology using deep Learning and Transfer Learning
Author Dataset Used Methodology Performance Measure
Z.Jiao et. al.
[38]
DDSM Deep features from
different layers and
SVM
Accuracy:96.7
J. Arevalo et. al
[39]
BCDR CNN and SVM AUC :0.826
B.Q.Huynh et al.
[40]
Private Dataset AlexNet and SVM AUC: 0.81
W.Sun et. al
[41]
In house FFDM Using CNN AUC 0.8818
R. K. Samala et. al.
[42]
DDSM Transfer learning in
Deep CNN
Single Task 0.78±0.02
Multitask 0.82±0.02
Lesion-based
0.76±0.01

N Antropova et
al.(2017)
[43]
FFDM
,Ultrasound
,DCE-MRI
CNN using VGG19,SVM AUC0.86,AUC:0.90,AUC:0
.89
N Dhungel et
al.(2017)
[44]
INBreast Pre train CNN using VGG
LeNet
AUC:0.91±0.12
V.Qiu et. al.(2017)
[45]
FFDM CNN with logistic
regression
AUC:0.69±0.044(1 –fold)
For entire data set
0.790±0.019
M.M Jadoon et al.
(2017)[46]
IRMA CNN,SVM AUC for CNN-WT 0.846
AUC for CNN-CT
0.855
X.Zhang et
al(2018)
[47]
Private Dataset AlexNet AUC for 2D 0.7274,3D
0.6632
H Chougrad, et. Al.
(2018) [48]
DDSM
,Inbreast
,BCDR
CNN using VGG-16,Res-
Net,InceptionV3
Accuracy
97.35%,95.50%,96.67%

Author Dataset Used Methodology Performance Measure
MA-AI-masni et al.
(2018)
[49]
DDSM ROI based CNN(YOLO) Accuracy 97%
D.Ribli et al.
(2018)
[50]
IN Breast, DDSM,
Private Dataset
Faster R-CNN AUC:0.95
H Wang et al.
(2018)
[51]
BCDR RNN AUC0.89
Shode Yu et al.
(2019)
[52]
BCDR GoogleLeNet,AlexNet,C
NN2 and CNN3,SVM
AUC0.82:0.88,AUC0.83,A
UC
Cai, hongmin et al
(2019)
[53]
Private Dataset Pre-trained using
AlexNet
Accuracy 0.8768

Summary Table
Model Description
Lenet-5 (1998) 7 level convolutional Network
AlexNet (2012) Top 5 errors from 26% to 15.5%
ZFNet (2013) Error rate 14.8
R-CNN,Fast R-
CNN,Mask R-CNN
(2014)
Sliding based(computation extensive and time consuming)
GoogLeNet (2014) Error rate 6.67
VGGNet (2014) 16 layer(Uniform Architecture)
YOLO (2015) Struggled with small objects)

•The images in DDSM are in an outdated image format with a bit depth of 12
or 16 bit per pixel. [54]
•BCDR-F03 contains gray scale TIFF(Tagged image File Format) bit depth of 8
bits per pixel.[54]
•Inbreast are saved in DICOM(Digital Imaging and Communications in
Medicine) format with 14 bit contrast resolution.[54]
•MIAS (Mammographic Image Analysis society)images are stored in 8 bits
per pixel in the PGM(Probabilistic graphical model) format. The database has
been reduced to a 200 micron pixel edge and padded/clipped so that the
image matrix is [1024,1024][54]

Flow of proposed Methodology
Classification(Benign/Malignant)
Solution for Overfitting
Classification of Image by applying CNN(VGG16)
Removal of the Artefacts/Noises/pectoral muscle
Orientation of the Image
Find MSE and PSNR to correct image.
Apply various types of the filters(Avg,Gaussian,Min,Max etc)
Add Noise to am image for all database presnt(S & P noise etc.)
Input Mammogram Image(322 Images)

Proposed Algorithm (B=Benign , M=Malignant)

Epoch=500(Overfitted Model)
Y axis= Loss, X axis= epoch
At y axis there is
Disturbance /Gap between
training and validation loss.
Because of this overfitting
the results are not Proper.
Need to solve Overfitting.

Epoch=500(Reduction of overfitting)
Y axis= Loss, X axis= epoch
The overfitting/Loss has now
reduced in(Y axis shows loss) .
As loss has been reduced from
y axis in much better manner
And independent variable
effect now reached to Zero.
Overfitting problem has been
solved and results are better.

Result
• Total Image taken for training ,validation and testing=115.
• Presently the accuracy have been received up to 92%
S. No. Algorithm/Model Results Accuracy
1 H Wang et al. (2018), RNN
[51]
AUC0.89
2 Shode Yu et al.,
GoogleLeNet,AlexNet,CNN2 and CNN3,SVM
(2019) [52]
AUC0.82:0.88,83
3 Cai, hongmin et al (2019), Pre-trained using
AlexNet,[53]
Accuracy 0.8768
4 Proposed Algorithm Accuracy =0.92

Future Scope
• To find the Exact boundary/Edge of Malignant
tumour.

REFERENCES:
[1] S. Ciatto, M.R.D. Turco, G. Risso, S. Catarzi, R. Bonaldi, V.Viterbo, P. Gnutti, B. Guglielmoni, L. Pinelli, A.
Pandiscia, F.Navarra, A. Lauria, R. Palmiero, P.L. Indovina, Comparison of standard reading and computer
aided detection (CAD) on a national proficiency test of screening mammography,European Journal of
Radiology 45 (2003) 135–138.
• [2] M.J. Homer, Mammographic Interpretation: A Practical Approach, McGraw-Hill, Boston, MA, 1997.
• [3] S.M. Kwok, R. Chandrasekhar, Y. Attikiouzel, M.T. Rickard,Automatic pectoral muscle segmentation on
mediolateral oblique view mammograms, IEEE Transactions on Medical Imaging 23 (September (9)) (2004)
1129–1140,doi:10.1109/TMI.2004.830529.
• [4]Petroudi, S.; Kadir, T.; Brady, M. Automatic classification of mammographic parenchymal patterns: a
statistical approach, Engineering in Medicine and Biology Society,2003. Proceedings of the 25th Annual
International Conference of the IEEE, vol. 1, pp. 798-801, 17–21 September,2003,
doi:10.1109/IEMBS.2003.1279885.
• [5]L. Vincent, P. Soille, Watersheds in digital spaces: an efficient algorithm based on immersion
simulations, IEEE Transactions on Pattern Analysis and Machine Intelligence 13 (June (6)) (1991) 583–598,
doi:10.1109/34.87344.

• 6] S. Tzikopoulos, H. Georgiou, M. Mavroforakis, N.Dimitropoulos, S. Theodoridis, A fully automated
complete segmentation scheme for mammograms, in: 16th International Conference on Digital Signal
Processing, 5–7 July 2009, 2009, pp. 1–6, doi:10.1109/ICDSP.2009.5201262.
• [7] M.J. Homer, Mammographic Interpretation: A Practical Approach, second ed., McGraw-Hill, New York,
1997.
• [8] http://www.imaginis.com/mammography/benefits-and-risks-of-mammography
• [9] R.E. Bird, T.W. Wallace, B.C. Yankaskas, Analysis of cancers missed at screening mammography,
Radiology 184 (3) (1992)613– 617.
• [10] R.G. Blanks, M.G. Wallis, S.M. Moss, A comparison of cancer detection rates achieved by breast cancer
screening programmes by number of readers, for one and two view mammography: Results from the UK
National Health Service Breast Screening Programme, The Journal of Medical Screening 5 (4) (1998) 195–
201.
• [11] T. Ojala, J. Nï?‘oeppi, O. Nevalainen, Accurate segmentation of the breast region from digitized
mammograms,Computerized Medical Imaging and Graphics 25 (January (1))(2001) 47–59, doi:
10.1016/S0895-6111(00)00036-7, ISSN:0895-6111.
• [12] R.B. Dubey, M. Hanmandlu, S.K. Gupta, A comparison of two methods for the segmentation of masses
in the digital mammograms, Computerized Medical Imaging and Graphics 34 (April (3)) (2010) 185–191,
doi: 10.1016/j.compmedimag.2009.09.002,ISSN: 0895-6111.
• [13] M. Rangaraj, Rangayyan, J. Fï?‘oebio, J.E. Ayres, Leo Desautels,A review of computer-aided diagnosis
of breast cancer:toward the detection of subtle signs, Journal of the Franklin Institute 344 (May–July (3–
4)) (2007) 312–348,
doi:10.1016/j.jfranklin.2006.09.003, ISSN: 0016-0032.

• [14] 8. Bhateja, V., Misra, M., Urooj, S., Lay-Ekuakille, A.: A robust polynomial filtering framework for
mammographic image enhancement from biomedical sensors. IEEE Sens. J. 13(11), 4147–4156 (2013)
• [15] K. Thangavel, M. Karnan, Computer aided diagnosis in digital mammograms: detection of
Microcalcifications by
• meta heuristis algorithms, ICGST-GVIP Journal 5 (7) (2005).
• [16] 13. Camilus, K., Govindan, V., Sathidevi, P.: Pectoral muscle identification in mammograms. J. Appl.
Clin. Med. Phys. North America 12(3), 215–230 (2011)
• [17] http://see.xidian.edu.cn/vipsl/database_Mammo.html
• [18] Sapate S., Talbar S. (2016) An Overview of Pectoral Muscle Extraction Algorithms Applied to Digital
Mammograms. In: Dey N., Bhateja V., Hassanien A. (eds) Medical Imaging in Clinical Applications. Studies
in Computational Intelligence, vol 651. Springer, Cham
• [19] Sapate, S.G., Talbar, S.N.: Pectoral muscle extraction using modified K-means algorithm for digital
mammograms. J. Med. Phys. (2016)
•
• [20] Karnan, M., Thangavel, K.: Automatic detection of the breast border and nipple position on digital
mammograms using genetic algorithm for asymmetry approach to detection of micro-calcifications.
Comput. Methods Programs Biomed. 87(1), 12–20 (2007)
•
• [21] Aroquiaraj, I.L., Thangavel, K.: Pectoral Muscles Suppression in Digital Mammograms using
Hybridization of Soft Computing Methods (2014).
•
• [22] Taghanaki, Saeid Asgari, et al. "Geometry Based Pectoral Muscle Segmentation from MLO
Mammogram Views." IEEE Transactions on Biomedical Engineering (2017).

• [23] TABALVANDANI, Nasibeh Saffari, and Domenec PUIG. "Feature Learning for Breast Tumour
Classification Using Bio-Inspired Optimization Algorithms." Recent Advances in Artificial Intelligence
Research and Development: Proceedings of the 20th International Conference of the Catalan Association
for Artificial Intelligence, Deltebre, Terres de L'Ebre, Spain, October 25-27, 2017. Vol. 300. IOS Press, 2017
• [24] Jeyasingh, Suganthi, and Malathi Veluchamy. "Modified Bat Algorithm for Feature Selection with the
Wisconsin Diagnosis Breast Cancer (WDBC) Dataset." Asian Pacific journal of cancer prevention: APJCP 18.5
(2017): 1257
•
• [25] Raja, N., et al. "Segmentation of Breast Thermal Images Using Kapur's Entropy and Hidden Markov
Random Field." Journal of Medical Imaging and Health Informatics 7.8 (2017): 1825-1829.
•
• [26] Khehra, Baljit Singh, and Amar Partap Singh Pharwaha. "Comparison of Genetic Algorithm, Particle
Swarm Optimization and Biogeography-based Optimization for Feature Selection to Classify Clusters of
Microcalcifications." Journal of The Institution of Engineers (India): Series B 98.2 (2017): 189-202.
• [27]Tsochatzidis, Lazaros, Lena Costaridou, and Ioannis Pratikakis. "Deep learning for breast cancer
diagnosis from mammograms—a comparative study." Journal of Imaging 5.3 (2019): 37.
• [28]https://www.google.com/search?q=breast+cancer+statistics+in+india+2020&tbm=isch&ved=2ahUKEw
jm4rS4663vAhUXSSsKHaCqD4UQ2-
cCegQIABAA&oq=breast+cancer+statistics+in+&gs_lcp=CgNpbWcQARgAMgIIADICCAAyAggAMgIIADICCAA
yAggAMgIIADICCAAyAggAMgQIABAYOgQIABBDUPvTAVis3QFg_ewBaABwAHgAgAGqAYgB-
QSSAQMwLjSYAQCgAQGqAQtnd3Mtd2l6LWltZ8ABAQ&sclient=img&ei=cvxMYObhKpeSrQGg1b6oCA&bih=
545&biw=1242&rlz=1C1CHBD_enIN924IN924#imgrc=sTQ_MBhHKKK_4M

[29] Smith, Robert D., and Mohandas K. Mallath. "History of the growing burden of cancer in India: from antiquity to the 21st
century." Journal of global oncology 5 (2019): 1-15.
[30] Duffy, Stephen W., et al. "Mammography screening reduces rates of advanced and fatal breast cancers: Results in 549,091
women." Cancer 126.13 (2020): 2971-2979.
[31] Jiang, Yulei, et al. "Improving breast cancer diagnosis with computer-aided diagnosis." Academic radiology 6.1 (1999): 22-33.
[32] Diz, Joana, Goreti Marreiros, and Alberto Freitas. "Using data mining techniques to support breast cancer diagnosis." New
Contributions in Information Systems and Technologies. Springer, Cham, 2015. 689-700.
[33] Zhang, Jing, James I. Silber, and Maciej A. Mazurowski. "Modeling false positive error making patterns in radiology trainees
for improved mammography education." Journal of biomedical informatics 54 (2015): 50-57.
[34] Malik, Bilal, et al. "Objective breast tissue image classification using Quantitative Transmission ultrasound
tomography." Scientific reports 6.1 (2016): 1-8.
[35] Muramatsu, Chisako, et al. "Breast mass classification on mammograms using radial local ternary patterns." Computers in
biology and medicine 72 (2016): 43-53.
[36] Sahran, Shahnorbanun, et al. "Machine learning methods for breast cancer diagnostic." Breast Cancer and Surgery.
IntechOpen, 2018.
[37] Peng, W., René V. Mayorga, and Esam MA Hussein. "An automated confirmatory system for analysis of
mammograms." Computer methods and programs in biomedicine 125 (2016): 134-144.
[38] Jiao, Zhicheng, et al. "A deep feature based framework for breast masses classification." Neurocomputing 197 (2016): 221-
231.
[39] Arevalo, John, et al. "Representation learning for mammography mass lesion classification with convolutional neural
networks." Computer methods and programs in biomedicine 127 (2016): 248-257.

[40] Huynh, Benjamin Q., Hui Li, and Maryellen L. Giger. "Digital mammographic tumor classification using
transfer learning from deep convolutional neural networks." Journal of Medical Imaging 3.3 (2016):
034501.
[41] Sun, Wenqing, et al. "Enhancing deep convolutional neural network scheme for breast cancer diagnosis
with unlabeled data." Computerized Medical Imaging and Graphics 57 (2017): 4-9.
[42] Samala, Ravi K., et al. "Multi-task transfer learning deep convolutional neural network: application to
computer-aided diagnosis of breast cancer on mammograms." Physics in Medicine & Biology 62.23 (2017):
8894.
[43] Antropova, Natalia, Benjamin Q. Huynh, and Maryellen L. Giger. "A deep feature fusion methodology for
breast cancer diagnosis demonstrated on three imaging modality datasets." Medical physics 44.10 (2017):
5162-5171.
[44] Dhungel, Neeraj, Gustavo Carneiro, and Andrew P. Bradley. "A deep learning approach for the analysis of
masses in mammograms with minimal user intervention." Medical image analysis 37 (2017): 114-128.
[45] Qiu, Yuchen, et al. "A new approach to develop computer-aided diagnosis scheme of breast mass
classification using deep learning technology." Journal of X-ray Science and Technology 25.5 (2017): 751-
763.
[46] Jadoon, M. Mohsin, et al. "Three-class mammogram classification based on descriptive CNN
features." BioMed research international 2017 (2017).
[47] Zhang, Xiaofei, et al. "Classification of whole mammogram and tomosynthesis images using deep
convolutional neural networks." IEEE transactions on nanobioscience 17.3 (2018): 237-242.
[48] Chougrad, Hiba, Hamid Zouaki, and Omar Alheyane. "Deep convolutional neural networks for breast
cancer screening." Computer methods and programs in biomedicine 157 (2018): 19-30.
[49] Al-Masni, Mohammed A., et al. "Simultaneous detection and classification of breast masses in digital
mammograms via a deep learning YOLO-based CAD system." Computer methods and programs in
biomedicine 157 (2018): 85-94.

[50] Ribli, Dezső, et al. "Detecting and classifying lesions in mammograms with deep learning." Scientific
reports 8.1 (2018): 1-7.
[51] Wang, Hongyu, et al. "Breast mass classification via deeply integrating the contextual information from
multi-view data." Pattern Recognition 80 (2018): 42-52.
[52] Yu, ShaoDe, et al. "Transferring deep neural networks for the differentiation of mammographic breast
lesions." Science China Technological Sciences 62.3 (2019): 441-447.
[53] Cai, Hongmin, et al. "Breast microcalcification diagnosis using deep convolutional neural network from
digital mammograms." Computational and mathematical methods in medicine 2019 (2019).
[54] Zou, Lian, et al. "A technical review of convolutional neural network-based mammographic breast cancer
diagnosis." Computational and mathematical methods in medicine 2019 (2019).
[55] Mutasa, Simukayi, Shawn Sun, and Richard Ha. "Understanding artificial intelligence based radiology
studies: What is overfitting?." Clinical imaging (2020).

Design of Modified Bio-Inspired Algorithm for Identification and Segmentation of Pectoral Muscle in Mammogram

More Related Content

What's hot (18)

Similar to Design of Modified Bio-Inspired Algorithm for Identification and Segmentation of Pectoral Muscle in Mammogram (20)

Recently uploaded (20)

Design of Modified Bio-Inspired Algorithm for Identification and Segmentation of Pectoral Muscle in Mammogram