Reproducibility of the lateralization index in functional magnetic resonance imaging across language tasks

Abstract A major challenge in computing the lateralization index (LI) of functional magnetic resonance imaging (fMRI) is the change in the index value according to the threshold applied and the task employed. To address this problem, we previously developed a threshold-free index called AveLI. AveLI is the average of the sub-LIs that are computed at all positive intensity values as the threshold. Thus, the algorithm has an intrinsic weighting on higher values due to the repeated involvement of the sub-LIs, and simultaneously also includes the lower intensities. While our previous study suggested the adequate reproducibility of AveLI between two tasks, we herein aimed to further investigate it using four different language tasks. We endeavored to include left-handed people to obtain a greater proportion of right hemisphere dominance in the 47 healthy participants analyzed. To characterize AveLI, we compared a total of six LI types including an index with a threshold and one with no threshold or weighting. We confirmed that AveLI had the highest reproducibility among all LI types examined. This was due to the use of the algorithm with weighting. A conventional LI with a threshold exhibited less reproducibility than AveLI, despite its ability to detect the left and right dominances to a similar extent as AveLI. This discrepancy suggested that AveLI encompassed an intrinsic lateralization that was hidden in the lower intensities, whereas the LI with the threshold omitted it to provide a task-specific lateralization.