Lerma-Usabiaga, G. 1 , Iglesias, J. E. 1 , Carreiras, M. 1, 2, 3 & Paz-Alonso, P. M. . 1
1 Basque Center on Cognition, Brain and Language (BCBL)
2 IKERBASQUE, Basque Foundation for Science, Bilbao, Spain
3 Departmento de Lengua Vasca y Comunicación, UPV/EHU, Bilbao, Spain
The human hippocampal formation is a crucial brain structure for memory and cognitive function that interacts extensively with distributed cortical and subcortical regions. Recent neuroimaging studies have found differences along the hippocampus longitudinal axis in terms of function, structure and connectivity with other regions, stressing the importance of improving the precision of the available segmentation methods typically used to divide this brain structure into anterior and posterior parts. In this regard, current segmentation conventions present two main sources of inaccuracies related to how separating planes along the longitudinal axis are chosen and how the in-scanner head position is corrected and equated across subjects before segmentation. These issues are typically addressed by manually aligning the brain for roll, pitch, and yaw rotations along the inter-hemispheric fissure, AC-PC line and orbits. Here, we propose an automated method based on estimating the longitudinal axis of the hippocampus with principal component analysis. The estimated direction is used to define the orientation of the separating planes, which removes the variability associated with the manual alignment of the in-scanner brain position. Our results show that this automatized procedure minimizes the error generated by the accumulation of manual operations while ensuring better reproducibility of results. This methodological improvement can potentially be used to improve the segmentation of other subcortical structures.