Hi,

I'm working on some images in order to extract the WMH but I've met some troubles while using your toolbox.

If I use the original images I obtain a probability map with all zero values, so the result is awful. Only when I multiply the image by a value of 10 or 20, I get some values different from zero in the final segmentation. I've been analyzing the code in MATLAB and I think that it may be caused due to the gray level of my image. Have you developed this toolbox for normalized images? or which is the range of gray levels in which the image is supposed to be?

I would be very grateful if you could help me,

Laura

Hello.

The toolbox was developed for native images i.e., no extra processing except bias correction is applied before segmenting the hyperintensities. 
A drawback of this is the aspect that for images with extra-bright layout (all voxels are somehow elevated more than in general), most of the regions on white matter will pop-up as detections in pmap.
Other users have also pointed out this issue with w2mhs. The only solution we suggest is adjusting the thresholds in WhyD_postproc. See the documentation.

Another option that I would suggest if the problem still persists is to change bias correction parameters in WhyD_preproc.m
This in certain cases 'dims-down' the whole image there by improving the segmentations. Thanks.

Hi Laura,

I happened to face the same issue right now and haven't solve the issue despite changes in cut values and trying different preproc bias values (60, 80, 100, 120,140). I was wondering if you managed to overcome this problem previously, and if so, may I ask how did you do it?

Best regards,
Jeremy