MRI training set for adaptive wavelet graph model.
The mat subdirectory contains 121 MRI reconstruction
images. The sup subdirectory contains the masks 
specifying the object support for the 121 images in mat.

The file 'cmdfile' in this directory can be used
as input file for the 'train' program in AWGM/bin/train.
The cmdfile uses 40 training images out of the 121 
total images in increments of 3 slices.

To use the cmdfile for training, first make sure that
the mri_train directory is installed in the same
directory as the AWGM package. For example, as

tomo/AWGM and tomo/mri_train.

Then, change to the AWGM directory.
From there, type 

bin/train ../mri_train/cmdfile

This will train the AWGM model on the 40 specified 
training images. After the training, you can delete
the (large) files AWGM/tree_dat/train* and 
AWGM/tree_dat/prune* . 
The results of the training will be stored in 
AWGM/tree_dat/clust* .

Note, that the images mri_train/mat/49.mat and 55.mat
are somewhat similar to the image 52.mat which I have
typically used for evaluation. However, leaving out
49.mat and 55.mat from the training 
does not result in a noticeable drop in reconstruction 
performance on 52.mat.

