#1
Nikolaus Weiskopf

7T and microstructures

#2
Yasmine Salman

50μm ex vivo MRI reveals distant layer-specific hippocampal degeneration in early AD

Background: Tau pathology early affects the medial temporal lobe (MTL), making it critical for understanding preclinical Alzheimer’s disease (AD). It is also the starting hub for TDP-43 pathology associated with limbic-predominant age-related TDP-43 encephalopathy (LATE-NC) and frontotemporal lobe dementia (FTLD). Although in vivo MRI can detect MTL atrophy, its sensitivity to subfield-specific changes and the relative contribution of tau and TDP-43 at early stages remain unclear. We aimed to identify MTL subregions affected by early tau pathology using high-resolution ex vivo MRI and histopathology and to assess the relative impact of tau and TDP-43 on MTL atrophy in individuals at early Braak NFT stages (≤II).

Methods: High-resolution (50μm) ex vivo 11.7T MRI scans of human MTL were acquired from 16 donors with early AD stages (Braak≤II), including three LATE cases with LATE-stage 1 (N=2) or 3 (N=1) and two FTLD cases. The anterior MTL was manually segmented into 12 subfields (Fig.1), including subiculum (SUB), CA1, CA2/3, dentate gyrus (DG), entorhinal cortex (ERC), and Brodmann area 35 (BA35). Region-of-interest landmarks were placed to estimate cortical thickness using a semi-automated approach. Counts of pTau/pTDP-43-positive neurons and total pathological particle burden of tau/TDP-43 were quantified in ERC/BA35 (Braak I ROI), CA1/SUB (Braak II ROI), and CA3/DG (Braak>III ROI; Fig.1).
Results: Comparisons between Braak 0/I (N=8) and Braak II (N=7), adjusted for age, sex, and co-pathologies (-synuclein/TDP-43), revealed significantly reduced CA1 pyramidal layer thickness in Braak II cases (Fig.2A; p<0.01). TDP-43–positive individuals (N=5) showed reduced thickness of the CA1 pyramidal layer and the CA2/3 stratum lacunosum-moleculare (SLRM) (Fig.2B). Partial Spearman correlations showed strong associations between CA1 pyramidal layer thickness and tau and TDP-43 burden (Fig.3). After adjusting for the other pathology, only the association with TDP-43 remained significant (R≤-0.61; p≤0.05), whereas tau showed a non-significant trend (R≤−0.47; p≤0.24).

By combining high-field ex vivo MRI with histopathology, we show that early AD-tau pathology is associated with CA1 pyramidal atrophy, identifying CA1 as a priority target for in vivo studies detecting preclinical AD. Importantly, TDP-43 substantially contributes to CA1 atrophy, underscoring the need to account for TDP-43 when interpreting early MTL structural changes.

#3
Milan Nemy

Deep learning-based cholinergic pathway imaging in Alzheimer’s disease

Background:The nucleus basalis of Meynert and its cholinergic white matter pathways are essential for cognition and are disproportionately affected in Alzheimer’s disease (AD). We investigated spatial and longitudinal patterns of cholinergic pathway degeneration across the AD continuum, with particular focus on preclinical and prodromal stages, and incorporated deep learning for automated pathway segmentation.

Methods: We analyzed diffusion MRI data from the DELCODE cohort in healthy controls (HC), subjective cognitive decline
(SCD), mild cognitive impairment (MCI), and AD dementia. Cholinergic pathways were assessed using our established tractbased framework, complemented by a custom encoder-decoder deep learning model that generated binary probability maps from slice-wise fiber-orientation features. Cross-sectional analyses included 402 participants (112 HC, 172 SCD, 66 MCI, 52 AD). Longitudinal analyses included 370 participants with at least two annual visits (mean maximum follow-up 2.80 years).
Mean diffusivity (MD) and fractional anisotropy (FA) were evaluated in cholinergic and non-cholinergic control pathways.
Voxel-wise cross-sectional analyses were corrected using permutation-based family-wise error correction. Longitudinal
trajectories were analyzed using linear mixed-effects models adjusted for age, sex, education, clinical stage, and scanning
protocol.

Results: Cross-sectionally, cholinergic pathways showed significantly increased MD and decreased FA already in SCD and in later disease stages (all p<0.001), whereas non-cholinergic control pathways showed no significant changes. Voxel-wise maps localized early abnormalities predominantly to the retrosplenial and posterior cingulate regions in SCD, with extension toward more anterior regions with increasing disease severity. Longitudinally, yearly MD change was non-significant in HC (-0.13 ± 0.17), but significant in SCD (0.70 ± 0.13), MCI (1.00 ± 0.22), and AD (2.05 ± 0.36) (all units ×10^-5 mm²/s/year; all p< 0.001). All between-group longitudinal differences were significant except SCD-MCI. The deep learning framework enabled automated, scalable segmentation of cholinergic pathways and supported robust extraction of diffusion biomarkers across the cohort.

Conclusions: Cholinergic white matter degeneration is detectable already in preclinical AD and shows stage-dependent progression across the disease continuum. Combining advanced diffusion MRI with deep learning-based segmentation enables reproducible assessment of this vulnerable system and supports cholinergic pathway imaging as a biomarker for early detection and disease monitoring in AD.

#4
Beatriz Padrela

Perfusion, Blood Brain Barrier and AD

Quantitative MRI gives new opportunities to investigate blood–brain barrier (BBB) function non-invasively. In this talk, I will
present a multi-echo arterial spin labeling (ASL) approach to estimate cerebral blood flow, arterial transit time, and water
exchange time (Tex), a surrogate marker of BBB permeability to water. I will present applications of this technique to study BBB in the context of AD and discuss current limitations and future methodological developments.