Difference between revisions of "CTSC Simon Warfield, CHB"

From NAMIC Wiki
Jump to: navigation, search
 
(29 intermediate revisions by the same user not shown)
Line 3: Line 3:
  
 
=Mission=
 
=Mission=
Description of big picture, goal(s) of project
+
Warfield_ALS_8-year-old-study. Description of big picture, goal(s) of project
  
=Use-Case Goals =
+
=Use-Case Goals =  
 +
We will approach this use-case in three distinct steps, including Basic Data Management, Query Formulation and Processing Support.
  
'''Step 1.'''
+
'''Step 1. Basic Data Management:'''
 +
* '''Step 1a.''' Upload retrospective data including MR DICOM and NRRD, associated .txt, .mat, etc. files.
 +
* '''Step 1b.''' Upload new acquisitions as part of data management pipeline (include EEG, MEG, data.) Clinical and Behavioral data considered later. Confirm with web GUI that data is present, organized and named appropriately.
  
'''Step 2.'''
+
'''Step 2. Query Formulation:'''  
 +
* Use web services API to formulate queries needed for study;
 +
* Confirm that query results match sets of data recovered locally for the same search.
 +
* Confirm that all queries required to support processing workflow work.
  
'''Step 3.'''
+
'''Step 3. Processing Workflow support:'''  
 +
* Implement and execute queries to support processing workflow, describe and upload processing results.
 +
* Ensure processing results are structured and named appropriately on repository, and queriable via web GUI and web services.
 +
 
 +
=Participants=
 +
* PI: Simon Warfield,
 +
* Co-Investigator: Neil Weisen (confirm)
 +
* Clinicians,
 +
* IT staff, Laura Alice (confirm)
  
 
=Outcome Metrics =
 
=Outcome Metrics =
 +
 +
'''Step 1: Data Management'''
 +
* Visual confirmation (via web GUI) that all data is present, organized and named appropriately
 +
* other?
 +
'''Step 2: Query Formulation'''
 +
*  Successful tests that responses to XNAT queries match search results on data in the local filesystem.
 +
* Query/Response should be efficient
 +
'''Step 3: Data Processing'''
 +
* Pipeline executes correctly
 +
* Pipeline execution not substantially longer than when all data is housed locally
 +
* other?
 +
'''Overall'''
 +
* Local disk space saved?
 +
* Data management more efficient?
 +
* Data management errors reduced?
 +
* Barriers to sharing data lowered?
 +
* Processing time reduced?
 +
* User experience improved?
  
 
=Fundamental Requirements =
 
=Fundamental Requirements =
  
=Participants=
+
* Excellent documentation
* PI, clinicians, IT staff, CT investigator, etc
+
* Example scripts to support custom query, custom schema extensions as required
* staff missing
+
* Data should be accessible 24/7
 +
* Guaranteed redundancy
 +
* Enough space to grow repository as required
 +
* Need a better client than current web GUI provides:
 +
** faster, present image data thumbnails (PACS-like)
 +
** cross platform, or web 2.0 compliant
 +
** robust image viewer (do not display jpegs)
 +
** must compete with functionality of OsiriX on MAC and other tools
 +
** would be good if it had interface to PACS.
 +
** recommend seeing approach taken by "Papers" software on mac.
 +
 
 +
=Outstanding Questions=
  
 
=Data=
 
=Data=
 +
Approximately 30 subjects currently, collection ongoing.
 +
 +
Retrospective data to manage in '''Step 1a''':
 +
* MR structural (DICOM + NRRD)
 +
* MR Diffusion (DICOM + NRRD)
 +
* MR Functional (DICOM + NRRD)
 +
* Protocol and associated data files (ascii text (.txt) and matlab .mat)
 +
 +
Ongoing acquisition data to manage in '''Step 1b''':
 +
* EEG
 +
* Clinical (paper intake, text)
 +
* Behavioral (paper intake, text)
 +
* MEG (Record + Photogrmty data, ascii text, European Data Format Plus (.edf))
  
 
=Workflows=
 
=Workflows=
  
 
==Current Data Management Process==
 
==Current Data Management Process==
 +
(schematics to come)
 +
 +
DICOM raw images are acquired at CHB and manually pushed to Research PACS. DCM4Chee currently provides the interface to Research PACS. A DICOM listener is set up, transfers data to local File System. Here a copy of the original data is made and converted to NRRD format. These new images are visually inspected in a QC step: in this step, the best representative dataset (Best_T1, Best_T2, and so on) is marked as the one to use for subsequent processing. The NRRD images are processed by a set of scripts which produce resulting datasets that describe diffusion characteristics, functional imaging activation statistics, etc. These results are also stored on the local file system. As part of the study, clinical and behavioral data is also collected; results are stored separately from the image data. Images managed on the filesystem are currently viewed with Osiris on Macs.
 +
 +
==Target Data Management Process (Step 1.)==
 +
'''Step 1''':
 +
 +
Develop an Image Management System for Warfield/CHB with which at least the following can be done:
 +
* Move images from CHB Research PACS to local disk for QC in usual way.
 +
* Perform QC assessment here in usual way.
 +
* '''Step 1a:'''
 +
** Describe & Import legacy MR, .txt and .mat data into XNAT instance;
 +
** Support custom tagging of imaging data to represent QC assessment + support subsequent processing.
 +
* '''Step 1b:''' Write scripts to execute upload of newly acquired data.
 +
** Potentially move data from CHB Research PACS directly to CHB XNAT
 +
** Support download for QC
 +
** Support tagging and upload post-QC.
  
==Target DataManagement Process (Step 1.)==
+
==Target Query Formulation for Processing (Step 2.) ==
 +
'''Step 2.''' Develop Query capabilities using scripted client calls to XNAT web services, such as:
  
==Target Query Formulation (Step 2.) ==
+
Show all subjectIDs in Project
 +
Show URI for subjectID=S & role=Best_T1
 +
Show URI for subjectID=S & role=Best_T2
 +
 
 +
* Scripting capabilities: Scripts need to
 +
** re-assign Role of other subject (modality) data if a new dataset for subject is tagged with Role=Best_(modality) (maintain only one best rating per modality per subject)
 +
** query and download data in a manner compatible with existing processing workflow.
  
 
==Target Processing Workflow (Step 3.)==
 
==Target Processing Workflow (Step 3.)==
 +
 +
'''Step 3:''' (schematic to come) Run processing locally, on cluster, etc.
 +
* Execute query/download script
 +
* Describe & upload processing results
 +
* Share images with clinical physicians (confirm)
 +
* Possibly export post-processed data back to research PACS
  
 
=Other Information=
 
=Other Information=

Latest revision as of 14:02, 30 September 2009

Home < CTSC Simon Warfield, CHB

Back to CTSC Imaging Informatics Initiative


Mission

Warfield_ALS_8-year-old-study. Description of big picture, goal(s) of project

Use-Case Goals

We will approach this use-case in three distinct steps, including Basic Data Management, Query Formulation and Processing Support.

Step 1. Basic Data Management:

  • Step 1a. Upload retrospective data including MR DICOM and NRRD, associated .txt, .mat, etc. files.
  • Step 1b. Upload new acquisitions as part of data management pipeline (include EEG, MEG, data.) Clinical and Behavioral data considered later. Confirm with web GUI that data is present, organized and named appropriately.

Step 2. Query Formulation:

  • Use web services API to formulate queries needed for study;
  • Confirm that query results match sets of data recovered locally for the same search.
  • Confirm that all queries required to support processing workflow work.

Step 3. Processing Workflow support:

  • Implement and execute queries to support processing workflow, describe and upload processing results.
  • Ensure processing results are structured and named appropriately on repository, and queriable via web GUI and web services.

Participants

  • PI: Simon Warfield,
  • Co-Investigator: Neil Weisen (confirm)
  • Clinicians,
  • IT staff, Laura Alice (confirm)

Outcome Metrics

Step 1: Data Management

  • Visual confirmation (via web GUI) that all data is present, organized and named appropriately
  • other?

Step 2: Query Formulation

  • Successful tests that responses to XNAT queries match search results on data in the local filesystem.
  • Query/Response should be efficient

Step 3: Data Processing

  • Pipeline executes correctly
  • Pipeline execution not substantially longer than when all data is housed locally
  • other?

Overall

  • Local disk space saved?
  • Data management more efficient?
  • Data management errors reduced?
  • Barriers to sharing data lowered?
  • Processing time reduced?
  • User experience improved?

Fundamental Requirements

  • Excellent documentation
  • Example scripts to support custom query, custom schema extensions as required
  • Data should be accessible 24/7
  • Guaranteed redundancy
  • Enough space to grow repository as required
  • Need a better client than current web GUI provides:
    • faster, present image data thumbnails (PACS-like)
    • cross platform, or web 2.0 compliant
    • robust image viewer (do not display jpegs)
    • must compete with functionality of OsiriX on MAC and other tools
    • would be good if it had interface to PACS.
    • recommend seeing approach taken by "Papers" software on mac.

Outstanding Questions

Data

Approximately 30 subjects currently, collection ongoing.

Retrospective data to manage in Step 1a:

  • MR structural (DICOM + NRRD)
  • MR Diffusion (DICOM + NRRD)
  • MR Functional (DICOM + NRRD)
  • Protocol and associated data files (ascii text (.txt) and matlab .mat)

Ongoing acquisition data to manage in Step 1b:

  • EEG
  • Clinical (paper intake, text)
  • Behavioral (paper intake, text)
  • MEG (Record + Photogrmty data, ascii text, European Data Format Plus (.edf))

Workflows

Current Data Management Process

(schematics to come)

DICOM raw images are acquired at CHB and manually pushed to Research PACS. DCM4Chee currently provides the interface to Research PACS. A DICOM listener is set up, transfers data to local File System. Here a copy of the original data is made and converted to NRRD format. These new images are visually inspected in a QC step: in this step, the best representative dataset (Best_T1, Best_T2, and so on) is marked as the one to use for subsequent processing. The NRRD images are processed by a set of scripts which produce resulting datasets that describe diffusion characteristics, functional imaging activation statistics, etc. These results are also stored on the local file system. As part of the study, clinical and behavioral data is also collected; results are stored separately from the image data. Images managed on the filesystem are currently viewed with Osiris on Macs.

Target Data Management Process (Step 1.)

Step 1:

Develop an Image Management System for Warfield/CHB with which at least the following can be done:

  • Move images from CHB Research PACS to local disk for QC in usual way.
  • Perform QC assessment here in usual way.
  • Step 1a:
    • Describe & Import legacy MR, .txt and .mat data into XNAT instance;
    • Support custom tagging of imaging data to represent QC assessment + support subsequent processing.
  • Step 1b: Write scripts to execute upload of newly acquired data.
    • Potentially move data from CHB Research PACS directly to CHB XNAT
    • Support download for QC
    • Support tagging and upload post-QC.

Target Query Formulation for Processing (Step 2.)

Step 2. Develop Query capabilities using scripted client calls to XNAT web services, such as:

Show all subjectIDs in Project
Show URI for subjectID=S & role=Best_T1
Show URI for subjectID=S & role=Best_T2
  • Scripting capabilities: Scripts need to
    • re-assign Role of other subject (modality) data if a new dataset for subject is tagged with Role=Best_(modality) (maintain only one best rating per modality per subject)
    • query and download data in a manner compatible with existing processing workflow.

Target Processing Workflow (Step 3.)

Step 3: (schematic to come) Run processing locally, on cluster, etc.

  • Execute query/download script
  • Describe & upload processing results
  • Share images with clinical physicians (confirm)
  • Possibly export post-processed data back to research PACS

Other Information