SlideShare a Scribd company logo
Overview of
Analysis Result Manager
         (ARM)
      Jeremy Magland

       27 October, 2011
  brainomics group @ Penn
Analysis Result Manager (ARM)
Motivations for ARM
O Motivation #1:
Visualization
I wanted to be able to
SEE my fMRI data,
intermediate files, and
processing results,
and to share these
with my collaborators.
Motivations for ARM
O Motivation #2:                            Bone-Water Study
                            Real-time Study Subject AP-1
Data Management              REAL_23            Localizer
                       Sample Session 1         Baseline, TR=300
Organize fMRI data     text     Raw fMRI series Follow-up 1, TR=300
                       About   Anatomic localizer
on a central server,   study Session 2    Bone-Water Study
                       Stuff REAL_36       Subject AP-1
and access it from     To fill Session 1     Localizer
                       sapace Session 2      Baseline, TR=300
anywhere (cloud                              Follow-up 1, TR=300

style)
Motivations for ARM
O Motivation #3:
Computing Power
Utilize the processing
power of server
computers (CPU
power, RAM,
automatic parallel
processing)
Motivations for ARM
O Motivation #4:
Provenance Tracking
Track the processing
steps required to
reproduce each
intermediate object and
processing result.
Advantages:
* reproducibility
* auto-non-redundant
processing
Challenge: Richly Interactive,
Web-based Data Visualization
O Browser-based
                           O RemoteWidgets:
  applications are very
  convenient but often       O Framework for
  have limited                developing
  functionality.
O Desktop applications        desktop
  can have rich               applications in Qt4
  functionality. However      with option to
  there are some cons:
   O software
                              deploy on browser.
     downloads/updates
   O OS incompatibility
   O Security risks
   O Etc.
RemoteWidgets: Overview
                           Seegull

O Write a GUI for the
  web using only one                 ARM
  programming
  language: Qt4/C++
O Rapid development       Pickles

  of highly interactive
  web applications.
O Direct access to                         Math Puppy

  server-side
  processes.
RemoteWidgets:
     One App, Two Access Options
                       Seegull in Web Browser

O Web Browser
  Interface (jQuery)
O Desktop Interface
  (thin client
  download)
                       Seegull on Desktop
RemoteWidgets: Hello, World!
RemoteWidgets:
      Server-Side Event Handling
O Handling GUI events is as easy as
 connecting Qt signals and slots on the
 server… no javascript coding needed.
RemoteWidgets: Idea
O Interactive, desktop-style visualization tools
  available from the convenience of a web
  browser.
Analysis Result Manager
             (ARM)
O Cloud-based
  approach to fMRI
  data management,
  visualization, and
  analysis.
O ARM is a
  RemoteWidgets
  application, and can
  therefore be
  accessed remotely
  via a web browser.
Tradeoff Between
    Scripting and GUI for fMRI Processing




O Scripting is way more powerful for research purposes.
  And, if you have a good script interface, you could always
  throw a GUI on top of it.
O But… some problems with Matlab-style scripting:
      O Need to manually shuffle data around in
           files/folders
      O Not easy to remember exactly which parameters
           were used >2 days ago
      O Run a 10-hour script, and discover the last line
           had a bug.
ARM: Script-Generated Workflows
With Robust Tracking of Provenance Data




   The multi-threaded processing     The script immediately sets up
   module then executes each         a workflow, creating all
   processors as soon as all input   intermediate and result objects
   objects have been defined.        ahead of the processing,
                                     storing detailed information
                                     about how objects are related
                                     via processes.
Processors are Generic Plugins
C++ Wrappers (for FSL, SPM, etc)




 Views are also plugins… (custom views of custom datatypes)
Live ARM Demonstration
O   Upload Scan Data                   O   View processing results and
O   Online view, change view               intermediate files.
    settings                           O   Show provenance (process
O   Super-efficient browsing of huge       history) of the output objects.
    data files.                        O   Explain that changing scan id’s
O   Browser access to same app             and even names of the
O   View example script                    processors causes no
                                           problems. Key assumption:
O   Running script does nothing            objects, once created, never
    because processing has already         change.
    completed
                                       O   Clean up by deleting results. (In
O   Change design function and re-         future, intermediate files will be
    run script – show how                  auto-removed after a period if
    processes are queued and               they are no longer linked to
    executed in proper order (No           results)
    redundant processing).
                                       O   In development – share a
                                           processing result by a button
                                           click.
Live Seegull Demonstration
O It is often desirable to SHARE results with others.
  But you don’t always want to give them access to a
  huge system. Furthermore, since ARM is highly
  dynamic, you want to be sure that a result, once
  shared, is available for ever after (snapshot of a
  result)
O Hence, Seegull
Summary and Future
O Share interactive views of data
O Taking full advantage of cloud technology
    to manage, process, visualize, compare
    fMRI data analyses
O   Script-generated workflows
O   Automatic provenance tracking
O   Convenient access via web browser
O   Import publicly available fMRI studies into
    system (powerful meta-analyses)
Especially Thank You to…
O Anna Rose Childress
O Felix Wehrli
O Charlie Epstein


And…
O   Benny Lam        O   John Detre     O Chris Tjoa
O   Oscar Bartra     O   Jesse Lupardus O Brian Avants
O   Kimberly Young   O   Ning Zhang
O   Dan Willard      O   Cheng Li
O   Rob Fabianski    O   Ze Wang

More Related Content

PDF
Parallel Programming in .NET
PDF
TMPA-2017: Distributed Analysis of the BMC Kind: Making It Fit the Tornado Su...
PPT
Migration To Multi Core - Parallel Programming Models
PPSX
Task Parallel Library Data Flows
PDF
Parallel program design
PPTX
C# Parallel programming
PPTX
From Python to smartphones: neural nets @ Saint-Gobain, François Sausset
PDF
More Data Science with Less Engineering: Machine Learning Infrastructure at N...
Parallel Programming in .NET
TMPA-2017: Distributed Analysis of the BMC Kind: Making It Fit the Tornado Su...
Migration To Multi Core - Parallel Programming Models
Task Parallel Library Data Flows
Parallel program design
C# Parallel programming
From Python to smartphones: neural nets @ Saint-Gobain, François Sausset
More Data Science with Less Engineering: Machine Learning Infrastructure at N...

What's hot (7)

PPTX
Async and parallel patterns and application design - TechDays2013 NL
PDF
GPU Computing for Data Science
PDF
Tensorflow on Android
PDF
TMPA-2017: Unity Application Testing Automation with Appium and Image Recogni...
PDF
Dark Silicon, Mobile Devices, and Possible Open-Source Solutions
PDF
STKO - A revolutionary toolkit for OpenSees
PPTX
IoT Supercharged: Complex event processing for MQTT with Eclipse technologies
Async and parallel patterns and application design - TechDays2013 NL
GPU Computing for Data Science
Tensorflow on Android
TMPA-2017: Unity Application Testing Automation with Appium and Image Recogni...
Dark Silicon, Mobile Devices, and Possible Open-Source Solutions
STKO - A revolutionary toolkit for OpenSees
IoT Supercharged: Complex event processing for MQTT with Eclipse technologies
Ad

Similar to Analysis Result Manager (ARM) (20)

DOCX
Remote Web Desk
PPT
Android overview
PPTX
Presentation1.2.pptx
PDF
OORPT Dynamic Analysis
PDF
Running Accurate, Scalable, and Reproducible Simulations of Distributed Syste...
PPTX
OS for AI: Elastic Microservices & the Next Gen of ML
ODP
Real time-collaborative-editor-presentation
PDF
Java Performance & Profiling
PPT
Presentation
PPTX
Running High-Speed Serverless with nuclio
PDF
Privacy-first in-browser Generative AI web apps: offline-ready, future-proof,...
PDF
Building and deploying LLM applications with Apache Airflow
PDF
Realizing the Promise of Portable Data Processing with Apache Beam
PPTX
IT TRENDS AND PERSPECTIVES 2016
DOCX
OBJECT ORIENTED ROGRAMMING With Question And Answer Full
PDF
Java Performance and Profiling
PDF
Google Developer Groups DEVELOPMENT WORKSHOP 2024.pdf
PDF
Overview of Modern Graph Analysis Tools
PDF
Software and the Concurrency Revolution : Notes
PPTX
Internship msc cs
Remote Web Desk
Android overview
Presentation1.2.pptx
OORPT Dynamic Analysis
Running Accurate, Scalable, and Reproducible Simulations of Distributed Syste...
OS for AI: Elastic Microservices & the Next Gen of ML
Real time-collaborative-editor-presentation
Java Performance & Profiling
Presentation
Running High-Speed Serverless with nuclio
Privacy-first in-browser Generative AI web apps: offline-ready, future-proof,...
Building and deploying LLM applications with Apache Airflow
Realizing the Promise of Portable Data Processing with Apache Beam
IT TRENDS AND PERSPECTIVES 2016
OBJECT ORIENTED ROGRAMMING With Question And Answer Full
Java Performance and Profiling
Google Developer Groups DEVELOPMENT WORKSHOP 2024.pdf
Overview of Modern Graph Analysis Tools
Software and the Concurrency Revolution : Notes
Internship msc cs
Ad

Recently uploaded (20)

PDF
Approach and Philosophy of On baking technology
PDF
Mushroom cultivation and it's methods.pdf
PDF
DP Operators-handbook-extract for the Mautical Institute
PPTX
Digital-Transformation-Roadmap-for-Companies.pptx
PDF
Transform Your ITIL® 4 & ITSM Strategy with AI in 2025.pdf
PDF
Encapsulation theory and applications.pdf
PDF
Getting Started with Data Integration: FME Form 101
PDF
WOOl fibre morphology and structure.pdf for textiles
PPTX
Group 1 Presentation -Planning and Decision Making .pptx
PPTX
TechTalks-8-2019-Service-Management-ITIL-Refresh-ITIL-4-Framework-Supports-Ou...
PDF
From MVP to Full-Scale Product A Startup’s Software Journey.pdf
PPTX
Tartificialntelligence_presentation.pptx
PDF
ENT215_Completing-a-large-scale-migration-and-modernization-with-AWS.pdf
PDF
DASA ADMISSION 2024_FirstRound_FirstRank_LastRank.pdf
PDF
Heart disease approach using modified random forest and particle swarm optimi...
PDF
MIND Revenue Release Quarter 2 2025 Press Release
PDF
Accuracy of neural networks in brain wave diagnosis of schizophrenia
PDF
gpt5_lecture_notes_comprehensive_20250812015547.pdf
PDF
Unlocking AI with Model Context Protocol (MCP)
PDF
Web App vs Mobile App What Should You Build First.pdf
Approach and Philosophy of On baking technology
Mushroom cultivation and it's methods.pdf
DP Operators-handbook-extract for the Mautical Institute
Digital-Transformation-Roadmap-for-Companies.pptx
Transform Your ITIL® 4 & ITSM Strategy with AI in 2025.pdf
Encapsulation theory and applications.pdf
Getting Started with Data Integration: FME Form 101
WOOl fibre morphology and structure.pdf for textiles
Group 1 Presentation -Planning and Decision Making .pptx
TechTalks-8-2019-Service-Management-ITIL-Refresh-ITIL-4-Framework-Supports-Ou...
From MVP to Full-Scale Product A Startup’s Software Journey.pdf
Tartificialntelligence_presentation.pptx
ENT215_Completing-a-large-scale-migration-and-modernization-with-AWS.pdf
DASA ADMISSION 2024_FirstRound_FirstRank_LastRank.pdf
Heart disease approach using modified random forest and particle swarm optimi...
MIND Revenue Release Quarter 2 2025 Press Release
Accuracy of neural networks in brain wave diagnosis of schizophrenia
gpt5_lecture_notes_comprehensive_20250812015547.pdf
Unlocking AI with Model Context Protocol (MCP)
Web App vs Mobile App What Should You Build First.pdf

Analysis Result Manager (ARM)

  • 1. Overview of Analysis Result Manager (ARM) Jeremy Magland 27 October, 2011 brainomics group @ Penn
  • 3. Motivations for ARM O Motivation #1: Visualization I wanted to be able to SEE my fMRI data, intermediate files, and processing results, and to share these with my collaborators.
  • 4. Motivations for ARM O Motivation #2: Bone-Water Study Real-time Study Subject AP-1 Data Management REAL_23 Localizer Sample Session 1 Baseline, TR=300 Organize fMRI data text Raw fMRI series Follow-up 1, TR=300 About Anatomic localizer on a central server, study Session 2 Bone-Water Study Stuff REAL_36 Subject AP-1 and access it from To fill Session 1 Localizer sapace Session 2 Baseline, TR=300 anywhere (cloud Follow-up 1, TR=300 style)
  • 5. Motivations for ARM O Motivation #3: Computing Power Utilize the processing power of server computers (CPU power, RAM, automatic parallel processing)
  • 6. Motivations for ARM O Motivation #4: Provenance Tracking Track the processing steps required to reproduce each intermediate object and processing result. Advantages: * reproducibility * auto-non-redundant processing
  • 7. Challenge: Richly Interactive, Web-based Data Visualization O Browser-based O RemoteWidgets: applications are very convenient but often O Framework for have limited developing functionality. O Desktop applications desktop can have rich applications in Qt4 functionality. However with option to there are some cons: O software deploy on browser. downloads/updates O OS incompatibility O Security risks O Etc.
  • 8. RemoteWidgets: Overview Seegull O Write a GUI for the web using only one ARM programming language: Qt4/C++ O Rapid development Pickles of highly interactive web applications. O Direct access to Math Puppy server-side processes.
  • 9. RemoteWidgets: One App, Two Access Options Seegull in Web Browser O Web Browser Interface (jQuery) O Desktop Interface (thin client download) Seegull on Desktop
  • 11. RemoteWidgets: Server-Side Event Handling O Handling GUI events is as easy as connecting Qt signals and slots on the server… no javascript coding needed.
  • 12. RemoteWidgets: Idea O Interactive, desktop-style visualization tools available from the convenience of a web browser.
  • 13. Analysis Result Manager (ARM) O Cloud-based approach to fMRI data management, visualization, and analysis. O ARM is a RemoteWidgets application, and can therefore be accessed remotely via a web browser.
  • 14. Tradeoff Between Scripting and GUI for fMRI Processing O Scripting is way more powerful for research purposes. And, if you have a good script interface, you could always throw a GUI on top of it. O But… some problems with Matlab-style scripting: O Need to manually shuffle data around in files/folders O Not easy to remember exactly which parameters were used >2 days ago O Run a 10-hour script, and discover the last line had a bug.
  • 15. ARM: Script-Generated Workflows With Robust Tracking of Provenance Data The multi-threaded processing The script immediately sets up module then executes each a workflow, creating all processors as soon as all input intermediate and result objects objects have been defined. ahead of the processing, storing detailed information about how objects are related via processes.
  • 16. Processors are Generic Plugins C++ Wrappers (for FSL, SPM, etc) Views are also plugins… (custom views of custom datatypes)
  • 17. Live ARM Demonstration O Upload Scan Data O View processing results and O Online view, change view intermediate files. settings O Show provenance (process O Super-efficient browsing of huge history) of the output objects. data files. O Explain that changing scan id’s O Browser access to same app and even names of the O View example script processors causes no problems. Key assumption: O Running script does nothing objects, once created, never because processing has already change. completed O Clean up by deleting results. (In O Change design function and re- future, intermediate files will be run script – show how auto-removed after a period if processes are queued and they are no longer linked to executed in proper order (No results) redundant processing). O In development – share a processing result by a button click.
  • 18. Live Seegull Demonstration O It is often desirable to SHARE results with others. But you don’t always want to give them access to a huge system. Furthermore, since ARM is highly dynamic, you want to be sure that a result, once shared, is available for ever after (snapshot of a result) O Hence, Seegull
  • 19. Summary and Future O Share interactive views of data O Taking full advantage of cloud technology to manage, process, visualize, compare fMRI data analyses O Script-generated workflows O Automatic provenance tracking O Convenient access via web browser O Import publicly available fMRI studies into system (powerful meta-analyses)
  • 20. Especially Thank You to… O Anna Rose Childress O Felix Wehrli O Charlie Epstein And… O Benny Lam O John Detre O Chris Tjoa O Oscar Bartra O Jesse Lupardus O Brian Avants O Kimberly Young O Ning Zhang O Dan Willard O Cheng Li O Rob Fabianski O Ze Wang