How to use matlab 'fitgmdist' for 3D brain image?

How can I fit a 3D-MRI brain data using 'fitgmdist' function provided by matlab. For example with matlab 'kmeans' function, I can cluster (stating the number of cluster) the 3D-image by flattening it and the output is always vector of the same size as the flattened input which I can then reshape back to 3D. How can I do the same using 'fitgmdist' function.

[Image Analyst]

So you think the distribution of gray levels is the sum of Gaussians? How many Gaussians do you think there are? Can you post a histogram of your image?

[Tolentee]

Yes! I think so. And I believe there should be about 3 Gaussians. Can't find an efficient way to display the 3D histogram.