search

markus-nilsson / md-dmri

MATLAB code for Multidimensional Diffusion MRI
Other
41 stars 34 forks source link

Master #9

Closed daniel-topgaard closed 7 years ago

daniel-topgaard commented 7 years ago

Improved plotting of output from dtd method

markus-nilsson commented 7 years ago

Nice work. Are there any scripts I would implement in a test suite to know that the new dtd things are working correctly?

I'm about to merge this changes into the master branch, but I omitted your abs(I) changes in mgui_contrast_load.m and in mgui_roi_update_panel.m, but added one real(I) for the display part... could you try if it works with your data now? abs(I) seems too aggressive with potential side effects when the data actually has negative values in it.

daniel-topgaard commented 7 years ago

Opening and analyzing the example data at https://github.com/daniel-topgaard/md-dmri-data https://github.com/daniel-topgaard/md-dmri-data in mgui is a simple test.

real(I) won’t work for the Bruker microimaging data which usually has some smooth phase variation across the image. Could you change to something like

if (any(imag(I) ~= 0)), I = abs(I); end

instead?

On 17 Mar 2017, at 22:16, Markus Nilsson notifications@github.com wrote:

Nice work. Are there any scripts I would implement in a test suite to know that the new dtd things are working correctly?

I'm about to merge this changes into the master branch, but I omitted your abs(I) changes in mgui_contrast_load.m and in mgui_roi_update_panel.m, but added one real(I) for the display part... could you try if it works with your data now? abs(I) seems too aggressive with potential side effects when the data actually has negative values in it.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/markus-nilsson/md-dmri/pull/9#issuecomment-287473357, or mute the thread https://github.com/notifications/unsubscribe-auth/AR1OU_C1fOVDuXH1ZpGPy3TDwF0-4r-yks5rmvgzgaJpZM4Ma9QX.

markus-nilsson commented 7 years ago

Wanted to keep the complex nature of the data as long as possible, to allow averaging over ROIs, for example. Now looking for this at all points where the code crashes with complex data. Seems to work with the soap example data.