Recent years have seen increased interest in measuring the subfields of the human hippocampal formation using MRI, with several research groups seeking to develop protocols and methods for delineating subfields in routine clinical resolution MRI data. However, progress in this area has been impeded somewhat by the significant amount of disagreement that exists between groups in the anatomical definition of the subfields, the strategies used to group smaller subfields into aggregate anatomical labels, and the rules used to delineate subfield boundaries in MRI data.
Furthermore, protocols designed for labeling different modalities for imaging the hippocampus, such as functional MRI, in vivo structural MRI, postmortem MRI microscopy, or histology, often demand different compromises between anatomical accuracy and repeatability, which leads to diverging protocols between the different communities within neuroimaging and between neuroimaging and neuroanatomy communities.
How to move forward
We believe that moving forward, it is critically important to quantitatively characterize the differences that exist between leading hippocampal subfield segmentation protocols, and where those differences are reconcilable, to move towards reducing them. With the emergence of a harmonized protocol for labeling the hippocampus as a whole (by Frisoni, Jack and colleagues), it is not unfeasible to imagine the subfield community one day coming together to develop a similar unified hippocampal subfield segmentation protocol.
With that goal in mind, we have gathered several meetings already. The Hippocampal Subfield Segmentation Summit (HS3) has been held at UC Davis (HS3.1), in San Diego CA as a satellite to the Society for Neuroscience meeting (HS3.2), UC Irvine (HS3.3), and more recently, in Chicago in late 2015. For more information on the conference and to register for upcoming conferences please click here.
- To bring together imaging researchers involved in designing subfield segmentation protocols, as well as neuroanatomists with MTL expertise, and computational algorithm developers;
- To examine differences in existing subfield segmentation protocols by comparing the segmentations of the same single-subject MRI dataset
- To establish whether a unified protocol (or protocols) is feasible and if so, to jointly develop an initial roadmap for making it happen;
- To begin work towards joint publications exploring the current state of the art, the major reasons for why protocols differ, and the strategies towards unification.
In addition to the organizers, this endeavor is only possible by contributions from a large number of laboratories who conduct high-resolution neuroimaging and conduct either manual, semi-automated or fully automated segmentations of hippocampal subfields. A list of the contributors and links to their respective labs can be found here.