Ultra High-Resolution 7T Ex Vivo MRI Analysis Reveals Structural-Pathological Associations in Alzheimer's Disease and Related Dementias

Core Concepts

Automated surface-based parcellation and vertex-wise analysis of ultra high-resolution 7T ex vivo MRI data from 82 brain donors with Alzheimer's disease and related dementias reveals significant associations between cortical thickness and neuropathological markers of disease.

Abstract

The study presents a computational pipeline that enables automated surface-based parcellation and vertex-wise analysis of ultra high-resolution (0.3 mm³) 7T ex vivo MRI data from 82 brain donors with Alzheimer's disease and related dementias (ADRD). Key highlights: The pipeline combines deep learning-based segmentation, topology correction, and surface-based modeling to parcellate the cortex using the Desikan-Killiany-Tourville (DKT) atlas in native subject space. Region-based analysis shows significant negative correlations between cortical thickness and neuropathological markers, including global amyloid-β, Braak staging, CERAD ratings, and regional tau pathology and neuronal loss in the medial temporal lobe. Vertex-wise analysis in template space reveals clusters of significant associations between cortical thickness and the neuropathological measures, particularly in the medial temporal lobe regions. This is the first large-scale study to perform structure-pathology correlation analysis using ultra high-resolution ex vivo MRI data, demonstrating the feasibility of applying in vivo neuroimaging analysis approaches to ex vivo datasets. The developed pipeline and dataset will be open-sourced to advance ex vivo MRI research on neurodegenerative diseases.

Stats

Postmortem interval (PMI) of 18.48 ± 13.60 hours and fixation time of 256.70 ± 280.48 days. Cohort included 41 female (age: 76.97 ± 9.70 years) and 41 male (age: 76.48 ± 11.67 years) patients with Alzheimer's Disease or related dementias.

Quotes

"Decades of neuroimaging research has yielded advanced computational frameworks for automated analysis of in vivo brain MRI, and tools such as FreeSurfer and Statistical Parametric Mapping (SPM) have been applied in a plethora of ADRD neuroimaging studies. However, these tools cannot be directly used on ultra-high resolution ex vivo MRI and there is very limited work on developing ex vivo MRI analysis methods for widespread use, which remains a challenge primarily due to the greater heterogeneity in scanning protocols of ex vivo MRI and increased complexity and imaging artifacts than in vivo MRI." "Crucially, the ability to map data from subject space to the DKT-template space based on surface correspondence is not limited to cortical thickness measures, and in future, our pipeline can be used to analyze features that take advantage of ultra-high resolution ex vivo MRI, such as mapping iron and myelin in cortex."

Key Insights Distilled From

Surface-based parcellation and vertex-wise analysis of ultra high-resolution ex vivo 7 tesla MRI in neurodegenerative diseases

by Pulkit Khand... at arxiv.org 03-29-2024

https://arxiv.org/pdf/2403.19497.pdf

Deeper Inquiries

How can the developed pipeline be extended to analyze other structural and functional features derived from ultra high-resolution ex vivo MRI, such as cortical myelination, iron content, and functional connectivity?

The developed pipeline for analyzing ultra high-resolution ex vivo MRI data can be extended to incorporate other structural and functional features by adapting the existing framework to accommodate the specific characteristics of these features. Here are some ways in which the pipeline can be extended: Cortical Myelination: To analyze cortical myelination, the pipeline can be modified to include imaging sequences or processing steps that specifically highlight myelin content in the brain. This may involve incorporating myelin-sensitive MRI contrasts or utilizing histological markers of myelin content to guide the analysis. Iron Content: For assessing iron content in the cortex, the pipeline can be adjusted to include sequences sensitive to iron deposition, such as susceptibility-weighted imaging. By integrating these sequences into the pipeline, researchers can quantify and map iron content across different brain regions. Functional Connectivity: To explore functional connectivity, the pipeline can be expanded to include resting-state functional MRI (rs-fMRI) data processing steps. This would involve extracting functional connectivity networks from the MRI data and integrating them with the structural information obtained from the high-resolution scans. Multi-Modal Integration: By integrating multiple modalities, such as diffusion tensor imaging (DTI) for white matter connectivity and PET imaging for molecular markers, the pipeline can provide a comprehensive view of the brain's structural and functional characteristics.

What are the potential limitations and sources of bias in using ex vivo MRI data for structure-pathology correlation studies compared to in vivo neuroimaging approaches?

Using ex vivo MRI data for structure-pathology correlation studies offers unique advantages but also presents certain limitations and sources of bias: Tissue Processing Artifacts: Postmortem tissue processing can introduce artifacts that may affect the quality of the MRI data, leading to potential distortions or signal irregularities that could impact the accuracy of structural measurements. Limited Temporal Information: Ex vivo MRI lacks the temporal dimension present in in vivo imaging, making it challenging to capture dynamic changes in brain structure and pathology over time. Selection Bias: The availability of ex vivo brain specimens is limited, leading to potential selection bias in the studied population. This could affect the generalizability of findings to broader populations. Histological Validation: While ex vivo MRI can provide high-resolution structural information, validating MRI findings with histological assessments is crucial. Variability in histological processing and staining techniques can introduce bias in the correlation studies. Loss of Functional Information: Ex vivo MRI focuses on structural imaging and may not capture functional aspects of the brain, limiting the ability to correlate structural changes with functional alterations seen in neurodegenerative diseases.

What insights can be gained by integrating the ultra high-resolution ex vivo MRI data with other modalities like histology and PET imaging to better understand the multifaceted neuropathological processes underlying Alzheimer's disease and related dementias?

Integrating ultra high-resolution ex vivo MRI data with histology and PET imaging can provide comprehensive insights into the neuropathological processes associated with Alzheimer's disease and related dementias: Correlative Analysis: By combining MRI data with histological assessments, researchers can validate structural findings with cellular and molecular markers, enhancing the understanding of the underlying pathology. Multi-Level Characterization: Integrating PET imaging data can offer information on molecular targets like amyloid and tau deposition, allowing for a multi-level characterization of disease pathology from macroscopic to molecular levels. Spatial Mapping: The integration of modalities enables precise spatial mapping of structural changes, protein aggregates, and metabolic activity, facilitating the identification of regional patterns of pathology in the brain. Pathological Staging: By correlating MRI features with histological markers and PET imaging results, researchers can establish links between structural alterations, proteinopathies, and disease progression stages, aiding in the development of staging models for neurodegenerative diseases. Therapeutic Targets: Insights from multimodal integration can help identify potential therapeutic targets by elucidating the relationships between structural changes, protein aggregation, and functional deficits, guiding the development of targeted interventions for Alzheimer's disease and related dementias.