Toward fMRI Group Identification Based on Brain Lateralization

This research presents a novel application of Lateralization Index (LI) in support of a decision making process for the classification of subjects based on their brain activation patterns using functional Magnetic Resonance Imaging (fMRI) datasets. The decision process considers the subject grouping based on additional spatial information provided by the LI behavior for each individual when calculated for specific Broca\´s and Wernicke\´s language areas. The presented results were obtained applying the LI concept to assess the activation pattern on both control and Localization-related epilepsy (LRE) subjects obtained during the execution of the language network oriented paradigm referred to as the "auditory description decision task" (ADDT). Upon assessing 114 datasets, activation was observed on 103 (90%) of them, while 11 (10%) showed no activation. Among these 103 datasets, 64 (62%) datasets were presumed as control data and 39 (38%) were presumed as LRE data. The data was obtained from 5 different hospitals using the online Web-based repository site (mri-cate.fiu.edu). Masks were used for temporal and lateral brain areas for the normal brain, and individual masks were used for 48 Brodmann areas (BA). A t-test yielded a P-value of 0.0151, which indicates a statistically significant difference in the mean of both groups. The LI was also calculated using both native and normal spaces for each subject, and in this case, no statistically significant difference between the two spaces was found. It is observed that the average brain activation intensity on the LRE subjects was higher than the one observed on the control population. On contrasting the LI percentages between control and LRE data (c%, e%), the following groups were identified: a) strong right lateralization: (0%, 18%), b) right lateralization: (2%, 10%), c) bilateral: (20%, 15%), d) left lateralization: (42%, 26%), e) strong left lateralization: (36%, 31%).