Multi-voxel pattern analysis of face and word encoding fMRI in people with temporal lobe epilepsy and healthy individuals.

Memory functional MRI (fMRI) has been used to explore cognitive processing in people with refractory temporal lobe epilepsy (TLE) to predict memory decline after anterior temporal lobe resection (ATLR). Traditional studies employed univariate analysis (UVA), focusing on isolated voxel activity in mesial temporal regions. By contrast, multivariate pattern analysis (MVPA), examines distributed activity patterns , offering deeper insight into neural networks supporting cognitive functions. MVPA allows precise mapping of individual memory encoding strategies that may influence surgical planning. Individual MVPA maps may be useful in the planning of surgical resection and trajectories for less-invasive procedures like laser interstitial thermotherapy (LiTT) to avoid critical brain regions. These may mitigate against the cognitive consequences of epilepsy surgery. We used MVPA to examine face and word encoding differences in healthy individuals and in people with TLE due to hippocampal sclerosis (HS). In healthy controls, medial temporal regions bilaterally supported face and word encoding. Lateral temporal regions were lateralized-right for faces and left for words. The fusiform gyri were involved in both tasks across all participants. HS patients showed altered encoding patterns, with reduced classifier accuracy (CA) in medial and lateral temporal regions. For face encoding, both left HS (LHS) and right HS (RHS) groups showed bilateral CA reductions in the parahippocampal gyrus (PHG), with additional hippocampal CA reduction in LHS. For word encoding, RHS patients showed CA reductions in right medial and bilateral lateral temporal regions, while LHS patients had reductions only in the left superior temporal gyrus (STG). These results highlight the bilateral network involved in memory encoding and the value of individualized cognitive mapping to prevent surgery-associated memory decline.