francopestilli / mba

Matlab Brain Anatomy. This is a set of routines for displaying and manipulating brain anatomy using MatLab.

Sunday coding 🌱

I am interested in Psychology, Neuroscience, Machine Learning, Informatics, Data Sharing, Governance and Data science. I work in a University because I believe in science and education. I like to think that by teaching young minds how to think critically and how to use cutting-edge tools and data, we can make the world a better place. It is not a simple task, but I believe we can improve society.

Connect with me on social media.

If you find any of my work helpful please give me a ⭐ at francopestilli

mba's Issues

matlabpool function should be removed

matlabpool() function in mbaDisplayConnectome.m causes a problem with new versions of Matlab (>2015a)
It is solved by just deleting lines 29 and 30 (see below)
% Handling parallel processing
poolwasopen=1; % if a matlabpool was open already we do not open nor close one
if (matlabpool('size') == 0), matlabpool open; poolwasopen=0; end

Reflects assumptions about streamlines fix recommended (and incoming)

mba/compute/mbaComputeFiberLengthDistribution.m

Line 30 in 1f06c4b

Lmm = cellfun('length',fg.fibers);

This computation assumes that (1) the nodes of streamlines are always 1mm spaced and (2) evenly spaced (such that each internode difference is equivalent).

Currently it is simply returning the number of nodes in all streamlines in an fg.

This is particularly a problem due to:

mba/compute/mbaRemoveFibersCoordsDistributionOutlier.m

Line 31 in 1f06c4b

if notDefined('numberOfNodes'), numberOfNodes=100; end

if notDefined('numberOfNodes'), numberOfNodes=100; end

which resamples all streamlines to the same number of nodes

Recommended fix:

streamsTotal=fg.fibers;
Lmm =[];
for istreamlines=1:length(fg.fibers)
Lmm(istreamlines)=sum(sqrt(sum(diff(streamsTotal{istreamlines},1,2).^2)));
end

(or some cellfun variant thereof to avoid the streamsTotal reassignment)

mbaGetSliceFromNifti vs. mbaGetResizedSliceImage

Hi Franco,

In mbaMakeImageFromNiftiSlice, are there any reasons to use mbaGetResizedSliceImage for extracting slice, instead of mbaGetSliceFromNifti?

line 39 in mbaGetSliceFromNifti.m
[img, x, y, z, plabe] = mbaGetSliceFromNifti(nifti,slice);

Sometimes we do not want to use resampling to 1mm (e.g. submilimeter data), and presumably it will be nice to have both options. If you think that this is a good idea, I'd be glad to do a pull request on the new version of mbaMakeImageFromNiftiSlice.m which can call both functions while we could keep using resizing as a default.

Missing adaptive parameter?

mba/display/mbaDisplayConnectome.m

Line 38 in 1f06c4b

% Low when many fibers are passed in, high otherwise.

This comment seems to suggest that the number of surfaces is scaled by the number of input streamlines, however, within this code no such scaling appears to be implemented.

More broadly, no such scaling appears to occur in any related wrapper.
https://github.com/francopestilli/mba/search?q=mbaDisplayConnectome

Is this a feature comment or a usage note/recommendation?

Dependency on MATLAB version...

Hi,

When I use MBA on MATLAB2015a environment, I got the following error.
Perhaps, "eml_scalexp_subsref" is missing in MATLAB 2015a. The code works perfectly at MATLAB 2012b, for example.

---

Error in normpdf (line 20)
xk = eml_scalexp_subsref(x,k);

Error in mbaComputeFibersOutliers (line 49)
maxoutDist = normpdf(maxDist); % Requires statistics toolbox

Error in s_nhp_segment_VOF_virtuallesion (line 42)
[fgsegment3, keepFascicle3] = mbaComputeFibersOutliers(fgsegment2,3,3,10);