CSVfilesDescription

For each cluster, the EM Clustering Module generates a set of CSV files. Let’s first define a couple of terms:

N_[clusterID] : number of significant trajectories that belong to cluster [clusterID].

L_[clusterID] : number of samples along the cluster centerline.

'-1': the missing data.

Now, the description of each file is as follows:

 [prefix]_posterior[clusterID].csv
 
  A single column (N_[clusterID] x 1) whose each element is the membership probability of a trajectory in that cluster.

 [prefix]_FA_cluster[clusterID].csv
  
  A matrix (N_[clusterID] x L_[clusterID]) whose each element(r,c) is the value of Fractional Anisotropy (FA) 
  for trajectory r corresponding to sample c of the cluster centerline.  

 [prefix]_MD_cluster[clusterID].csv
  
  A matrix (N_[clusterID] x L_[clusterID]) whose each element(r,c) is the value of Mean Diffusivity (MD) 
  for trajectory r corresponding to sample c of the cluster centerline.  

 [prefix]_Par_cluster[clusterID].csv
  
  A matrix (N_[clusterID] x L_[clusterID]) whose each element(r,c) is the value of parallel diffusivity (largest Eigen value of the tensor matrix)
  for trajectory r corresponding to sample c of the cluster centerline.  

 [prefix]_Per_cluster[clusterID].csv
  
  A matrix (N_[clusterID] x L_[clusterID]) whose each element(r,c) is the value of perpendicular diffusivity (mean value of the two Eigen values 
  - other than the largest one - of the tensor matrix) for trajectory r corresponding to sample c of the cluster centerline.

Matlab scripts here can be used to generate plots of diffusion parameters along each cluster.