Test systems data preparation

[sheryjoe]

Data are needed to be prepared for all below test systems to evaluate and validate ShapeWorks tools.

In house:

• [x] left atrium (LA)
• [x] left atrium appendage (LAA) with LA
• [x] right ventricle (RV)
• [x] biventricle - missing images
• [x] left atrium, right atrium, and septum (LA-RA-S)
• [x] ankle
• [x] scapula
• [x] humerus
• [x] shoulder (scapula and humerus)
• [x] femur
• [x] hip (femur + pelvis) (cartilage?)

Public:

• [x] prostate - does not have images
• [x] carma-LA
• [x] cell/nuclear
• [x] cardiac LA
• [x] hippocampus
• [x] prostate
• [x] spleen
• [x] Stanford Knee
• [x] LV/RV
• [x] 4D cardiac
• [x] Total segmentator

Each test system include the following. Data should be organized based on the portal datasets requirements.

• imaging data (CT or MRI, etc)
• anatomy segmentation (surface meshes and/or binary segmentations)
• data-specific annotations (landmarks, constraints) (optional)
• metadata (pathology vs controls, outcomes, demographic, age, gender, etc)
• ShapeWorks spreadsheet project file (only the data sheet).

[jadie1]

We are keeping track of this via this spreadsheet: https://docs.google.com/spreadsheets/d/1MpYoPxEbv0IsqWzHvmDUB6QE3L_n__Dlegsk-NLrLYY/edit?ts=60c7815a#gid=0

[iyerkrithika21]

In the last Cardiology meeting Jake had a question regarding the size of the datasets. Do we have any specifications for that ? And do we need to include pathological and controls in all datasets or just controls would be enough?

[sheryjoe]

Moving this to 6.3 but we have to finish this by December max regardless of the release date.

[cchriste]

Moving this to 6.3 but we have to finish this by December max regardless of the release date.

Looks really close. Might it have been completed?

[jadie1]

I am just finishing up organizing the prostate dataset

[iyerkrithika21]

I need to finish organizing the biventricle data.

[jadie1]

Datasets that need a license file:

• Atria (LA-RA-S)
• Left Atrium
• Left Atrium Appendage
• Biventricle
• Right Ventricle
• Ankle

I will ask Amy about the ankle license. @akenmorris and @sheryjoe who should I ask about the cardiac licenses?

[jadie1]

Potential public test system: https://www2.childmind.org/webmail/908232/621532556/ab06ec718e8753cf11f86c9c7a8da595abbd6e12bf69f0e2a59ab80f3243a0be

[jadie1]

Note the data folder has changed from /usr/sci/data/SSM-Data/ to CHPC storage automounted at /usr/sci/datanew/SSM-Data/

[jadie1]

I have added the TotalSegmentator dataset to /usr/sci/datanew/SSM-Data/Public_Data/Totalsegmentator_dataset/. This dataset contains 1204 subjects with 104 shape classes (i.e., bones, organs, muscles, etc). There are 76,888 non-empty segmentations, and of those 49,216 do not lie on image boundary/are not cutoff, these have been marked as "complete".

The test systems spreadsheet has been updated and I created a spreadsheet for the entire dataset and complete data: /usr/sci/datanew/SSM-Data/Public_Data/Totalsegmentator_dataset/all_shapes.xlsx and /usr/sci/datanew/SSM-Data/Public_Data/Totalsegmentator_dataset/complete_shapes.xlsx

I've also added json files with the shape counts for the available and complete segmentations: /usr/sci/datanew/SSM-Data/Public_Data/Totalsegmentator_dataset/all_shape_counts.json and /usr/sci/datanew/SSM-Data/Public_Data/Totalsegmentator_dataset/complete_shape_counts.json