IronTract Challenge - Round II - How to Participate

Transcript

1. How to participate IronTract Challenge Round II
2. Step 1: Register to the Challenge https://irontract.mgh.harvard.edu/register/ Request challenge-specific access codes for your team Submitted only once per team
3. Step 2: Register on the QMENTA Platform https://client.qmenta.com/
4. Step 2: Register on the QMENTA Platform
5. Step 3: Sign in the QMENTA Platform
6. Step 3: Sign in the QMENTA Platform
7. Step 3: Sign in the QMENTA Platform
8. Step 4: Download the training data
9. Step 5: Run your favorite tractography method! ROUND I PARTICIPANTS ROUND II PARTICIPANTS Use the same orientation reconstruction and tractography methods that you used in Round I Use any orientation reconstruction and tractography method of your choice
10. Step 5: Run your favorite tractography method! ROUND I PARTICIPANTS ROUND II PARTICIPANTS Use the same orientation reconstruction and tractography methods that you used in Round I Use any orientation reconstruction and tractography method of your choice Both previous and new participants must use the harmonized pre-processing and seed region
11. Step 6: Postprocessing of the training data •Post-processing 1: spatial filtering, from EPFL team compute_density_volumes_EPFL.py. Trk files use the script as is Tract Volumes apply the first step (length thresholding) on the fly in your own tractography code, and you can then use the rest of the script on your output volumes. • Post-processing 2: anatomical ROIs, from VUMC team compute_density_volumes_VUMC.py + two inclusion masks Trk files use the script as is Tract Volumes apply inclusion masks on the fly in your tractography code, and you can then use the rest of the script on your output volumes You will receive a link to access the following scripts upon registering to the challenge Email us if you have any question!
12. Step 6: Postprocessing of the training data •Post-processing 1: spatial filtering, from EPFL team compute_density_volumes_EPFL.py. Trk files use the script as is Tract Volumes apply the first step (length thresholding) on the fly in your own tractography code, and you can then use the rest of the script on your output volumes. • Post-processing 2: anatomical ROIs, from VUMC team compute_density_volumes_VUMC.py + two inclusion masks Trk files use the script as is Tract Volumes apply inclusion masks on the fly in your tractography code, and you can then use the rest of the script on your output volumes You will receive a link to access the following scripts upon registering to the challenge Include all the NIfTI volumes produced by this method at all the different thresholds in a single zip file. Submit results separately for each of these two post-processing approaches
13. Step 6: Upload the training data
14. Step 6: Upload the training data
15. Step 6: Upload the training data
16. Step 6: Upload the training data
17. Step 6: Upload the training data: check results Max 10 Uploads If you need more, contact us
18. Step 7: Download the validation data
19. Step 8: Postprocessing of the validation data •Post-processing 1: spatial filtering, from EPFL team compute_density_volumes_EPFL.py. Trk files use the script as is Tract Volumes apply the first step (length thresholding) on the fly in your own tractography code, and you can then use the rest of the script on your output volumes. • Post-processing 2: anatomical ROIs, from VUMC team compute_density_volumes_VUMC.py + two inclusion masks Trk files use the script as is Tract Volumes apply inclusion masks on the fly in your tractography code, and you can then use the rest of the script on your output volumes You will receive a link to access the following scripts upon registering to the challenge Include all the NIfTI volumes produced by this method at all the different thresholds in a single zip file. 1-2 pages description of your methods. PDF included in the zip file of your validation results. - must include any information that one would need to replicate your results independently. Submit results separately for each of these two post-processing approaches
20. Step 8: Upload the validation data
21. Step 8: Upload the validation data
22. By November 8th you need to upload: 1.Results from the training data. 1.Results from the validation data. 1.A 1-2 page description of your methods. - PDF included in the zip file of your validation results. - must include any information that one would need to replicate your results independently. All of the above must be the same for the training and validation data. SHARE YOUR CODE + Finalize your submission
23. Subscribe to the mailing list: ayendiki@mgh.havard.edu cmaffei@mgh.Harvard.edu Questions?