The medial temporal lobe (MTL) is a necessary component in a variety of memory functions, as well as the locus of structural change in aging, Alzheimer's disease (AD), schizophrenia, and other conditions. The distinct subregions composing the MTL, including various subfields of the hippocampus, have been implicated in different memory subsystems, and shown to be differentially affected in normal aging and AD. The ability to reliably and efficiently detect these subregions using in vivo neuroimaging would therefore be of great potential value for both basic neuroscience and clinical research. Such a procedure will provide critical insights into the function and structure of the MTL in the living human brain, and how it is affected in normal aging. It is also an important step in the quest for sensitive, non-invasive biomarkers for early diagnosis and treatment evaluation in AD. The limited resolution of typical MRI scans has traditionally been a major hindrance in imaging studies of the MTL, forcing investigators to treat the hippocampus and surrounding structures as a single entity. Substantial developments in MR data acquisition technology, however, have started to yield images that show anatomical features of the MTL at an unprecedented level of detail, providing the basis for fine-scaled functional and morphological analyses of individual subregions of the MTL. MRI studies of the MTL at the subfield level are currently not widely performed. This is because they require a combination of deep MRI know-how, neuroanatomical expertise, and staffing resources available only at a select few specialized sites. In order to make MRI studies of the MTL at the subfield level more widely accessible, the overall goal of this project is to develop and validate a broadly applicable set of computational tools to automatically segment a multitude of MTL subregions from in vivo MRI images. Specifically, given the extreme versatility of MRI and the lack of standard acquisition protocols for imaging the MTL, we will build tools that can robustly analyze scans of various image resolutions and tissue contrasts. Towards this end, we aim to (1) use manual delineations in ultra-high resolution MRI scans to derive computational models that make predictions about the relative position and shape of MTL subregions, (2) based on these models and on a model of the MRI imaging process, develop and validate a Bayesian framework for fully-automated MTL subregion segmentation in ultra-high resolution MRI scans, and (3) develop and validate such a framework for lower resolution images acquired on systems in more widespread use, by explicitly accounting for the partial volume effect where several structures contribute to form the intensity within a single voxel. In order to disseminate the developed techniques and atlases to the scientific community, we plan to integrate them into an open source package that we will make freely available as part of the FreeSurfer environment.

Public Health Relevance

The ability to reliably measure subtle degenerative changes in small medial temporal lobe (MTL) substructures through in vivo MRI would be an important step towards early diagnosis and staging of Alzheimer's disease, and towards monitoring therapeutic interventions. Such a procedure could also provide unprecedented insights into changes in the MTL structure in the living human brain that accompany normal aging, a crucial clinical and neuroscientific objective as the MTL is a brain region known to be critical in the human memory system.

Agency
National Institute of Health (NIH)
Institute
National Institute of Biomedical Imaging and Bioengineering (NIBIB)
Type
Research Project (R01)
Project #
5R01EB013565-04
Application #
8642178
Study Section
Biomedical Imaging Technology Study Section (BMIT)
Program Officer
Pai, Vinay Manjunath
Project Start
2011-04-01
Project End
2015-03-31
Budget Start
2014-04-01
Budget End
2015-03-31
Support Year
4
Fiscal Year
2014
Total Cost
$458,975
Indirect Cost
$199,667
Name
Massachusetts General Hospital
Department
Type
DUNS #
073130411
City
Boston
State
MA
Country
United States
Zip Code
02199
Saygin, Z M; Kliemann, D; Iglesias, J E et al. (2017) High-resolution magnetic resonance imaging reveals nuclei of the human amygdala: manual segmentation to automatic atlas. Neuroimage 155:370-382
Menze, Bjoern H; Van Leemput, Koen; Lashkari, Danial et al. (2016) A Generative Probabilistic Model and Discriminative Extensions for Brain Lesion Segmentation--With Application to Tumor and Stroke. IEEE Trans Med Imaging 35:933-46
Cantero, Jose L; Iglesias, Juan E; Van Leemput, Koen et al. (2016) Regional Hippocampal Atrophy and Higher Levels of Plasma Amyloid-Beta Are Associated With Subjective Memory Complaints in Nondemented Elderly Subjects. J Gerontol A Biol Sci Med Sci 71:1210-5
Menze, Bjoern H; Jakab, Andras; Bauer, Stefan et al. (2015) The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS). IEEE Trans Med Imaging 34:1993-2024
Iglesias, Juan Eugenio; Augustinack, Jean C; Nguyen, Khoa et al. (2015) A computational atlas of the hippocampal formation using ex vivo, ultra-high resolution MRI: Application to adaptive segmentation of in vivo MRI. Neuroimage 115:117-37
Yushkevich, Paul A; Amaral, Robert S C; Augustinack, Jean C et al. (2015) Quantitative comparison of 21 protocols for labeling hippocampal subfields and parahippocampal subregions in in vivo MRI: towards a harmonized segmentation protocol. Neuroimage 111:526-41
Iglesias, Juan Eugenio; Van Leemput, Koen; Bhatt, Priyanka et al. (2015) Bayesian segmentation of brainstem structures in MRI. Neuroimage 113:184-95
Iglesias, Juan Eugenio; Sabuncu, Mert Rory; Aganj, Iman et al. (2015) An algorithm for optimal fusion of atlases with different labeling protocols. Neuroimage 106:451-63
Van Leemput, Koen; Sabuncu, Mert R (2014) A cautionary analysis of STAPLE using direct inference of segmentation truth. Med Image Comput Comput Assist Interv 17:398-406
Iglesias, Juan Eugenio; Sabuncu, Mert Rory; Van Leemput, Koen (2013) A probabilistic, non-parametric framework for inter-modality label fusion. Med Image Comput Comput Assist Interv 16:576-83

Showing the most recent 10 out of 22 publications