search

CCI-Bonn / HD-GLIO

Automated deep-learning based brain tumor segmentation on MRI
Apache License 2.0
26 stars 7 forks source link

D #11

Open skgtohn opened 1 year ago

skgtohn commented 1 year ago

Dear Fabian,

Hi Fabian.

I have recently used HD-GLIO to segment tumors in my dataset inputs (t1, t1-ce, t2, t2-flair), and the segmentation worked well. However, I noticed that the resulting segmented tumor has binary outputs (either gray or white, depending on the opacity settings). I was wondering if this is intended for HD-GLIO since I want to sublabel the tumors from the segmented tumor.

Furthermore, I would like to know if it is possible to only segment each modality image instead of generating a single segmented tumor from the four image modalities.

Thank you for your time and assistance.

Best regards, Curtis

FabianIsensee commented 1 year ago

Hey Curtis, HD-GLIO should predict two separate labels for edema and enhancing tumor. It does not segment necrosis. If only one label is returned, then this means that only one class (probably edema) was detected in that particular case. Best, Fabian