Hello,
The French registry on Multiple Sclerosis (OFSEP) aims at processing its High Definition Cohort (5000 patients) to extract MS lesions from brain MRI.
I am in charge of comparing MS lesion segmentation methods* to select the one which will be used for this task. Methods able to differentiate new lesions from existing ones from longitudinal data are particularly interesting, but all methods will be considered.
Here is the curated list that I established from the literature:
- nnUNet
- Multiple Sclerosis Lesion Segmentation with Tiramisu and 2.5D Stacked Slices
- Spatio temporal MS Lesion Segmentation
- MDGRU (still to be tested)
- A cross-sectional method from the Empenn team at Inria
- A longitudinal method detecting new lesions developed at Empenn from the paper A fully convolutional neural network for new T2-w lesion detection in multiple sclerosis and nnUNet.
I would like to make sure that I did not miss any important method for this task; please tell me if you think of a MS lesion segmentation method able to compete with those listed above.
Also, I am looking for similar initiatives elsewere in the world; the broader goal being to identify new biomarkers for multiple sclerosis.
Note: OFSEP is not only interested in publicly available methods, but is also ready to use commercial methods as long as an agreement can be found between OFSEP and the method owners.
Thanks!