Temporal Lobe Epilepsy with Mesial-Sclerosis: The Development of a Valid, Reliable, Harmonized Segmentation Protocol for Hippocampal Subfields
Abstract number :
V.060
Submission category :
5. Neuro Imaging / 5A. Structural Imaging
Year :
2021
Submission ID :
1825722
Source :
www.aesnet.org
Presentation date :
12/9/2021 12:00:00 PM
Published date :
Nov 22, 2021, 06:50 AM
Authors :
Susanne Mueller, Dr. med. - University of California, San Francisco; Ana Daugherty, PhD - Wayne State University; Renaud La Joie, PhD - University of California, San Francisco; Davic Berron, PhD - Lund University; Arnold Bakker, PhD - John Hopkins University; Naftali Raz, PhD - Wayne State University; Craig Stark, PhD - University of California, Irvine; Paul Yushkevich, PhD - University of Pennsylvania; Lei Wang, PhD - Ohio State University; Valerie Carr, PhD - San Jose State University; Rosanna Olsen, PhD - Baycrest Health Sciences; Laura Wisse, PhD - Lund University
Rationale: Histopathological studies have shown that hippocampal atrophy in temporal lobe epilepsy with mesio-temporal sclerosis (TLE-MTS) is not diffuse but shows distinct atrophy patterns, (1) cornu ammonis (CA1) atrophy, (2) atrophy of the dentate gyrus (DG) or end folium sclerosis, (3) atrophy of CA1 and DG, and (4) global hippocampal atrophy. There is also evidence that some atrophy pattern are associated with a better chance of post-surgical seizure-freedom than others. High resolution images at 3T and higher depict some details of the internal structure of the hippocampus in vivo. This allowed for the development of several manual and automated subfield labeling schemes with the aim to identify the atrophy pattern in vivo. The results of these studies regarding prediction of the post-surgical outcome have been inconsistent though. One of the reasons for this are the substantial differences in how different labs define the subfield boundaries. The Hippocampal Subfields Group (HSG) is an international group seeking to address this issue by developing a histologically-valid, reliable, and freely available segmentation protocol for high-resolution T2-weighted 3T MRI (http://www.hippocampalsubfields.com).
Methods: Our workflow consists of four steps: 1) collecting histology samples labeled by multiple expert neuroanatomists to form a novel reference dataset to guide the development of the MRI segmentation protocol (cf. Fig 1), 2) developing boundary definitions for each segment of the hippocampus, (head, body, and tail) and MTL cortices on MRI, 3) assessing HSG community agreement with boundary rules using online questionnaires and 4) testing reliability of the protocol definitions
Results: For the hippocampal body and head, we have developed a preliminary subfield segmentation protocol (i.e. completed steps 1-2). For the boundaries of the body with surrounding regions, 29 labs reached consensus for all rules (step 3). We are now administering additional questionnaires for assessing agreement of the hippocampal body and head subregion boundary rules. Upon completion, reliability testing of the protocol should begin in 2021 for the body (step 4).
Conclusions: Once completed, the harmonized protocol can be used for automated and manual labeling. This will significantly facilitate the testing of new automated subfield labeling approaches, their optimization for TLE, the comparison of the results across labs and ultimately result in a robust way to identify atrophy pattern associated with good post-surgical outcome.
Funding: Please list any funding that was received in support of this abstract.: EU Joint Programme—Neurodegenerative Disease Research (JPND) project.
Neuro Imaging