Would it be possible to share ADNI dataloader files

[grant-jpg]

Excellent work! I'm wonder whether it would be possible to share ADNI dataloader python files.

A question here, are masks needed to construct a dataset? Based on my interpretation, the dataset just needs to contain image and class label. But the code constructing biobank dataset seems to require masks provided. If I misunderstood your work, please let me know.

[ecr05]

Cher is no longer working on the project so not sure she can help. I have Cc'd Helena in case she can answer any of the questions, but I don't think we have access to any more of the historical code.

Sorry

Emma

Sent from Outlook for Androidhttps://aka.ms/AAb9ysg

---

From: grant @.> Sent: Monday, April 3, 2023 1:15:14 PM To: CherBass/ICAM @.> Cc: Subscribed @.***> Subject: [CherBass/ICAM] Would it be possible to share ADNI dataloader files (Issue #7)

Excellent work! I'm wonder whether it would be possible to share ADNI dataloader python files.

A question here, are masks needed to construct a dataset? Based on my interpretation, the dataset just needs to contain image and class label. But the code constructing biobank dataset seems to require masks provided. If I misunderstood your work, please let me know.

— Reply to this email directly, view it on GitHubhttps://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FCherBass%2FICAM%2Fissues%2F7&data=05%7C01%7Cemma.robinson%40kcl.ac.uk%7Ccefcc48cc4b6455ebeef08db343d1564%7C8370cf1416f34c16b83c724071654356%7C0%7C0%7C638161209172338568%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=iNXfiOSDljud4rHQv0i7tsqfCX%2BhJ6SpDqPv5%2FZxByw%3D&reserved=0, or unsubscribehttps://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAD7GXPFJMRUYY7GBA4W4ZULW7K5NFANCNFSM6AAAAAAWRHNPSY&data=05%7C01%7Cemma.robinson%40kcl.ac.uk%7Ccefcc48cc4b6455ebeef08db343d1564%7C8370cf1416f34c16b83c724071654356%7C0%7C0%7C638161209172338568%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=ezeah4OYM6%2Fcmotp3qYRE2jWOsJdDWAoKK4%2FQs6x5dQ%3D&reserved=0. You are receiving this because you are subscribed to this thread.Message ID: @.***>

[helenass97]

Hi, thank you for your comment and sorry for the very late reply.

Masks are not needed to construct the datasets (they are only optional)... the image and class label are the only essentials as you mentioned - but if you can/want to use masks for the datasets you are also able to use the cross-correlation option ( --cross-cor in options.py ) for example.

And I've attached here the original python script for the ADNI dataloader (might need some slight modifications to work). Thank you!

adni_dataloader.py.zip