This thesis demonstrates the use of the bootstrap resampling technique considering temporal dependency in the fMRI data to determine the reliability and confidence interval of fMRI parameters. Traditionally, the test-retest method has been used to reliably detect active voxels in the fMRI image of the brain, which is based on repetitive experimentation. The main concern with the test-retest method is the reproducibility of data over these multiple repetitions. Fatigue, habituation, motion artifacts, and repositioning errors are few of the factors, which can affect the reproducibility of data.

The conventional bootstrap resampling technique is based on the assumption that the dataset is independent and identically distributed over time. However, studies have shown temporal dependency in the fMRI images of the brain acquired from subjects in the resting phase. This study demonstrates the use of the bootstrap resampling technique, incorporating the criterion of temporal dependency in the fMRI data set, to detect reliable active voxels in the fMRI images acquired during a task activated motor paradigm, where the subject is instructed to perform bilateral finger tapping.

The results of the study showed that the active regions detected using the bootstrap resampling technique considering temporal dependency in the fMRI data were more reliable than the active regions detected using the bootstrap resampling technique without considering any temporal dependency in the fMRI data.