Feature selection technique for time-series fMRI data of schizophrenia patients

Related Research Article: Mean deviation based identification of activated voxels from time-series fMRI data of schizophrenia patients,  F1000Research.

Cite the article as "Chatterjee I. Mean deviation based identification of activated voxels from time-series fMRI data of schizophrenia patients. F1000Research 2018, 7:1615 (doi: 10.12688/f1000research.16405.1)"

Cite the software as " CHATTERJEE, INDRANATH. (2018, September 29). Feature selection technique for time-series fMRI data of schizophrenia patients (Version v1.0.0). Zenodo. http://doi.org/10.5281/zenodo.1438539 "

By: Indranath Chattterjee

Department of Computer Science, University of Delhi, Delhi-110007,

E-mail: indranath.cs.du@gmail.com

Run the code in the following order:

Step 1. Run the 'script_dataload.m' locating the proper directory containing all the runs of the preprocessed time-series fMRI data.

Step 2. Run the 'intersection_union_script.m'

Step 3. If want to classify with SVM, run 'classification_SVM.m', and if with ELM then run 'classification_ELM.m'

The MAT-file named 'Contrast_AudOdd_Dev_Std_15T_2D.mat' contains the data for 34 healthy subjects (first 34) and 34 schizophrenia patients (next 34). These are the contrast maps obtained from GLM analysis of time-series fMRI data for 4 runs of Auditory-oddball task. Each row contains the 153594 voxels covering the whole brain for each subject. The 4 runs of each subject's data are averaged.

1. This program will fetch the time-series fMRI data of each of the 4 runs of the Auditory-oddball (AUD) task for each of the 34 healthy subjects and 34 schizophrenia patients.

2. This program will try to find the changes in the activation pattern within a particular voxel along its time-series with the help of mean-deviation based analysis.

3. Using the simple set operations, the index of the relevant brain voxels will be obtained.

4. This index file can be fed as input to the classifier as a feature set during building the classification model.