functional brain clustering

Here we apply hierarchical clustering on both row and column (biclustering) of the data matrix. Each column represents a vector capturing the responses of class-specific distinguishability measure of 8 classes.

From the results, the bottle and scissor are merged first, saying that they are most similar. House and human face are very stand-out from others. The interesting pair is cat and scrambled image.

Red means great similarity --> Blue means small similarity. What I like to see more is which part of the brain corresponds to which class. But it seems that clustering on voxels makes it hard to see the maximally-response area in the brain, so I'm trying to group them and plot on the brain.

Now, if we cluster the voxels according to the 8D vector, what would we get?

I will need help to put the cluster on the T1 brain.