Postmortem ex vivo MRI at 7 tesla

Pulkit Khandelwal,

Michael Tran Duong,

Shokufeh Sadaghiani,

Sydney Lim,

Amanda Denning,

Eunice Chung,

Sadhana Ravikumar,

Sanaz Arezoumandan,

Claire Peterson,

Madigan Bedard,

Noah Capp,

Ranjit Ittyerah,

Elyse Migdal,

Grace Choi,

Emily Kopp,

Bridget Loja,

Eusha Hasan,

Jiacheng Li,

Alejandra Bahena,

Karthik Prabhakaran,

Gabor Mizsei,

Marianna Gabrielyan,

Theresa Schuck,

Winifred Trotman,

John Robinson,

Daniel Ohm,

Edward B. Lee,

John Q. Trojanowski,

Corey McMillan,

Murray Grossman,

David J. Irwin,

John Detre,

M. Dylan Tisdall,

Sandhitsu R. Das,

Laura E.M. Wisse,

David A. Wolk,

Paul A. Yushkevich

University of Pennsylvania
Paper arXiv Video Code Docker
exvivo_image

Abstract

Ex vivo MRI of the brain provides remarkable advantages over in vivo MRI for visualizing and characterizing detailed neuroanatomy, and helps to link microscale histology studies with morphometric measurements. However, automated segmentation methods for brain mapping in ex vivo MRI are not well developed, primarily due to limited availability of labeled datasets, and heterogeneity in scanner hardware and acquisition protocols. In this work, we present a high resolution of 135 ex vivo post-mortem human brain tissue specimens scanned on a 7T whole-body MRI scanner. We developed a deep learning pipeline to segment the cortical mantle by benchmarking the performance of nine deep neural architectures. We then segment four subcortical structures (caudate, putamen, globus pallidus, and thalamus), white matter hyperintensities, and the normal appearing white matter. We show excellent generalizing capabilities across whole brain hemispheres in different specimens, and also on unseen images acquired at different magnetic field strengths and different imaging sequence. We then compute volumetric and localized cortical thickness measurements across key regions, a link them with semi-quantitative neuropathological ratings. Our code, containerized executables, and the processed datasets are publicly available.

exvivo_image

exvivo_image

Teaser video

Surface-based parcelaltion and vertex-wise analysis

We adapted and developed a fast and easy-to-use automated surface-based pipeline to parcellate, for the first time, ultra high-resolution ex vivo brain tissue at the native subjectspace resolution using the Desikan-Killiany-Tourville (DKT) brain atlas. This allows us to perform vertex-wise analysis in the template space and thereby link morphometry measures with pathology measurements derived from histology.

exvivo_freesurfer_image
Ex vivo MRI and parcellations. Sagittal, axial and coronal viewing planes of ex vivo MRI at 0.3mm3 resolution for two subjects (A and B) with corresponding DKT volumetric segmentations and surface-based parcellations on pial and inflated surfaces for the medial and lateral views in subject space resolution. Our method is able to correctly delineate the brain even in regions where the MR signal contrast is low in the anterior and the posterior brain MRI due to artifacts in acquisition protocol. Bottom row: the segmentations and parcellations on the corresponding 3T in vivo MRI of subject B which shows that the developed pipeline correctly map the labels to similar brain locations in ex vivo MRI.

exvivo_freesurfer_results_image
Template-space vertex-wise morphometry-pathology correlations. Vertex-wise group analysis was performed to fit a generalized linear model (GLM). Shown are the statistical map (t-statistics) of the correlation between cortical thickness (mm) and with global ratings of amyloid-β, Braak staging, CERAD, and semiquantitative ratings of the medial temporal lobe (MTL) tau pathology and neuronal loss, the with age, sex and postmortem interval (PMI) as covariates across all 82 subjects. The clusters outlined in white indicate regions significant correlations (p<0.05) were observed after FWER correction for multiple comparisons.

Acknowledgements

We gratefully acknowledge the tissue donors and their families. We also thank all the staff at the Center for Neurodegenerative Research (University of Pennsylvania) for performing the autopsies and making the tissue available for this project. This work was supported in part by the National Institute of Health Grants: P30 AG072979, R01 AG056014, RF1 AG069474, R01 AG054519, P01 AG017586, U19 AG062418.

BibTeX

@article{khandelwal2023automated,
  title={Automated deep learning segmentation of high-resolution 7 T ex vivo MRI for quantitative analysis of structure-pathology correlations in neurodegenerative diseases},
  author={Khandelwal, Pulkit and Duong, Michael Tran and Sadaghiani, Shokufeh and Lim, Sydney and Denning, Amanda and Chung, Eunice and Ravikumar, Sadhana and Arezoumandan, Sanaz and Peterson, Claire and Bedard, Madigan and others},
  journal={arXiv preprint arXiv:2303.12237},
  year={2023}
}