Aug 2026· Scientific Reports· Vol 16· 0 citations· 56 references
Medicine
TL;DR
A scalable, open-source, deep-learning approach to quantify NFT burden in digital whole slide images (WSIs) of post-mortem human brain tissue and openly releases this multi-institution deep-learning pipeline to provide detailed NFT spatial distribution and morphology analysis capability at a scale otherwise infeasible by manual assessment.
Abstract
Accumulation of abnormal tau protein into neurofibrillary tangles (NFTs) is a pathologic hallmark of Alzheimer disease (AD). Accurate detection of NFTs in tissue samples can reveal relationships with clinical, demographic, and genetic features through deep phenotyping. However, expert manual analysis is time-consuming, subject to observer variability, and cannot handle the data amounts generated by modern imaging. We present a scalable, open-source, deep-learning approach to quantify NFT burden in digital whole slide images (WSIs) of post-mortem human brain tissue. To achieve this, we developed a method to generate detailed NFT boundaries directly from single-point-per-NFT annotations. We then trained a semantic segmentation model on 45 annotated 2400 μm by 1200 μm regions of interest (ROIs) selected from 15 unique temporal cortex WSIs of AD cases from three institutions (University of California (UC)-Davis, UC-San Diego, and Columbia University). Segmenting NFTs at the single-pixel level, the model achieved an area under the receiver operating characteristic of 0.832 and an F1 of 0.527 (196-fold over random) on a held-out test set of 664 NFTs from 20 ROIs (7 WSIs). We compared this to deep object detection, which achieved comparable but coarser-grained performance that was 60% faster. The segmentation and object detection models correlated well with expert semi-quantitative scores at the whole-slide level (Spearman’s rho ρ = 0.654 (p = 6.50e-5) and ρ = 0.513 (p = 3.18e-3), respectively). We openly release this multi-institution deep-learning pipeline to provide detailed NFT spatial distribution and morphology analysis capability at a scale otherwise infeasible by manual assessment.
Introduction Cerebral amyloid angiopathy (CAA) is characterized by amyloid-beta deposition in cortical and leptomeningeal vessels and associated with cognitive impairment and hemorrhage. Current neuropathological assessments rely on semiquantitative grading and lack vessel-level resolution and scalability. Existing computational pathology approaches also fail to capture individual vessel morphology and spatial amyloid distribution across whole-slide images (WSIs). To address this gap, we developed a deep learning framework for reproducible, quantitative analysis of CAA in WSIs. Methods We analyzed 20 postmortem brain tissue sections from the frontal (n = 10) and occipital cortices (n = 10) of 10 individuals with Alzheimer’s disease pathology obtained from the University of Pittsburgh Alzheimer’s Disease Research Center, which served as the internal development cohort. An independent external cohort consisted of 10 sections (5 frontal and 5 occipital samples) from 5 individuals obtained from the University of Kentucky Alzheimer’s Disease Research Center. We trained and compared three semantic segmentation architectures, a standard U-Net, a dual-attention residual U-Net (DA-ResUNet), and a Swin Transformer-based U-Net (Swin-UNet), using the internal development cohort with slide-level five-fold cross-validation. All models were evaluated on the independent external cohort to assess generalization under domain shift. Based on segmentation performance and computational efficiency, we selected one architecture to generate whole-slide composite segmentation masks for vessel walls, amyloid deposits, and tissue compartments. These masks were subsequently used for deterministic vessel detection, morphometric measurements, and quantification of vascular and perivascular amyloid features through post-processing analysis. Results All three architectures achieved high segmentation accuracy on the internal cohort, with Dice scores above 90% across vessel walls, amyloid deposits, gray matter, and leptomeninges. The Swin-UNet showed marginally higher performance for vessel segmentation, whereas the DA-ResUNet provided more balanced accuracy and computational efficiency and was selected for downstream analysis. External cohort evaluation demonstrated robust generalization, with attention-enhanced models outperforming the standard U-Net under domain shift. Using the selected model, the pipeline reliably detected valid vessels, excluded non-vascular artifacts, and enabled deterministic extraction of vessel morphometry, vascular and perivascular amyloid burden, and identification of circumferential CAA involvement at the vessel level. Discussion This framework provides a scalable, interpretable solution for vessel-level CAA analysis, supporting robust geometric and spatial characterization of cerebrovascular pathology and enabling future integration with clinical and genetic studies. Beyond CAA, the modular design allows extension to other vascular pathologies, including arteriolosclerosis, in WSIs, facilitating broader investigation of cerebrovascular disease mechanisms.
H. Tahmasebidehkordi, A. Bahramy, Dana R. Julian et al.· bioRxiv· 0 citations
Advances in image registration and machine learning have recently enabled volumetric analysis of postmortem brain tissue from conventional photographs of coronal slabs, which are routinely collected in brain banks and neuropathology laboratories around the world. One caveat of this methodology is the requirement of segmentation of the tissue from the background and out-of-slice tissue in photographs, which currently requires laborious manual intervention. Manual delineation is a bottleneck in this process and poses challenges in scalability, and resources, restricting adoption of these methods. In this article, we present a deep learning model to automate this process. The automatic segmentation tool relies on a U-Net architecture that was trained with a combination of 1,414 manually segmented images of both fixed and fresh tissue, from specimens with varying diagnoses, photographed at two different sites. Automated model predictions on a subset of photographs not seen in training were analyzed to estimate performance compared to manual labels, including both inter- and intra-rater variability. Our model achieved a median Dice score over 0.98, mean surface distance under 0.4 mm, and 95% Hausdorff distance under 1.60 mm, which approaches inter-/intra-rater levels. Our tool is publicly available at surfer.nmr.mgh.harvard.edu/fswiki/PhotoTools and training data is available at https://zenodo.org/records/20647553.
Jonathan Williams Ramirez, Dina Zemlyanker, Lucas J. Deden-Binder et al.· PLoS ONE· 0 citations
BACKGROUND
Enlarged perivascular spaces (ePVS) are a marker of cerebral small vessel disease, potentially reflecting reduced waste clearance. Because manual quantification is unfeasible in large datasets, we developed and evaluated an automated tool.
METHODS
Detection Of Regions of Enlarged perivascular Spaces (DORES), a 3D nnU-Net-based deep learning algorithm was developed for ePVS segmentation using T1-weighted and fluid-attenuated inversion recovery magnetic resonance imaging (MRI). DORES was developed in two stages: an initial model trained on 35 manually segmented scans and a final model on 1460 pseudo-labeled sessions from the Vanderbilt Memory and Aging Project (VMAP). A subset of VMAP participants with 3 T brain MRI underwent whole-brain manual ePVS tracing (n = 35, 73 ± 9 years, 51% male) and visual rating (n = 388, 71 ± 8 years, 54% male) by a neuroradiologist. DORES was evaluated and compared against three other segmentation tools using Dice and F1 scores, absolute volume and element differences, correlation, and agreement. External validation used an Alzheimer's Disease Neuroimaging Initiative 3 subset with manual tracings (ADNI3, n = 18, 73 ± 9 years, 67% female).
RESULTS
DORES achieved Dice scores of 0.61 ± 0.16 (white matter) and 0.72 ± 0.08 (basal ganglia) in VMAP, with strong correlations and agreement for ePVS count and volume. Performances modestly declined in ADNI3 across algorithms. Scanner-stratified analyses showed stronger correlations for Philips versus Siemens images in the basal ganglia, indicating scanner-dependent differences in measurement consistency.
CONCLUSIONS
DORES provides a multimodal nnU-Net-based pipeline for ePVS segmentation in older adults. The model demonstrates robust within-cohort performance and reasonable external validity, though scanner-related effects limit application across sites.
James D. LeFevre, W. Robb, Dandan Liu et al.· Magnetic Resonance Imaging· 0 citations
Fine analysis of the spatial distribution and morphology of Aβ plaques is crucial for understanding the pathological progression of Alzheimer's disease (AD). However, at the whole-brain scale, the enormous number of plaques, wide size range, diffuse morphologies, and complex imaging background pose challenges that existing digitization methods often fail to address comprehensively. To this end, this study developed a Generalized Plaque Digitization Framework (GPDigit). The framework adopts a two-step strategy of detection followed by segmentation: first, a customized object detection network achieves precise spatial localization of plaques; second, adaptive foreground signal segmentation is performed within local regions. This design circumvents the difficulty of global threshold selection and the high annotation cost of segmentation networks. GPDigit supports both 2D and 3D data scenarios, ensures detection accuracy through targeted feature extraction mechanisms, and significantly improves training data preparation efficiency with a self-developed annotation tool and multiple data augmentation strategies. Experimental results demonstrate that GPDigit achieves satisfactory digitization performance under challenging conditions such as dense plaque distribution, severe background interference, and weak signals. Application to whole-brain plaque analysis in 5xFAD mice across multiple key ages revealed, at the fine brain-region scale, the spatiotemporal heterogeneity of plaque density and load development. This study provides a systematic solution for plaque digitization in neuropathological mesoscopic high-resolution images and offers a practical tool to advance AD pathology research.
Precise brain segmentation is fundamental for quantitative neuroimaging analysis. However, most existing methods lack generalization across the human lifespan and diverse imaging modalities, limiting their utility for Comprehensive Brain Segmentation (CBS) (i.e., tissue segmentation, parcellation, and lesion labeling). To address this, we propose BrainSeg, a novel unified framework, for CBS by using large-scale datasets spanning the entire lifespan, with adaptability to diverse uni- and multimodal input scenarios without the need for retraining or finetuning. Comprehensive experiments are conducted on lifespan data ranging from 14 gestational weeks to 100 years of age, consisting of 45,998 multimodal scans from 26 datasets, which are further augmented by our proposed synthesis strategy. Systematic validation and in-depth analysis demonstrate that our BrainSeg can achieve state-of-the-art performance across all three core CBS tasks, with the averaged Dice ratios reaching up to 96.94% for tissue segmentation, 94.25% for brain parcellation, and 91.06% for lesion labeling in the internal validations. It maintains similarly high accuracy in external validations, with averaged Dice ratios achieving 94.01% for tissue segmentation, and 91.20% for brain parcellation, underscoring its robustness and generalizability across diverse conditions. In summary, BrainSeg serves as a versatile foundation tool, providing flexible and reliable analysis for large-scale neuroimaging studies.
Shijie Huang, Zifeng Lian, Dengqiang Jia et al.· npj Digital Medicine· 1 citation
The rising global burden of cerebrovascular disease, propelled by an aging population, highlights the inherent limitations of conventional, labor-intensive diagnostic paradigms. In the context of time-sensitive stroke management, variability in image interpretation and the high rate of misclassification, particularly during the assessment of transient ischemic attack (TIA), underscore the urgent need for more consistent and efficient diagnostic solutions. Artificial intelligence (AI), particularly deep learning (DL), offers a transformative pathway by automating the analysis of complex neurovascular imaging. Here, we conduct a comprehensive examination of how DL is revolutionizing stroke-related image analysis, moving beyond general assertions of potential to discuss specific technical implementations. We systematically detail the evolution from traditional segmentation algorithms to advanced deep learning architectures—such as U-Net, DeepMedic, and their variants—in performing critical tasks. These tasks encompass the automated segmentation of intracranial and extracranial (carotid) arteries, the quantification of stenosis and plaque burden, and the hemodynamic assessment of vascular lesions across modalities including MRA, CTA, and DSA. By synthesizing landmark studies, our analysis delineates three core aspects: the technological trajectory of DL models in achieving expert-level accuracy in vascular feature extraction in controlled studies; the clinical translation of these tools into diagnostic, prognostic, or therapeutic procedural planning workflows; and the persistent challenges and future directions, including data standardization, model generalizability, and multimodal integration. This review posits that DL represents not merely an assistive technology but a foundational cornerstone for the next generation of precision cerebrovascular medicine. It holds the potential to bridge critical gaps in diagnostic speed, objectivity, and accessibility, provided its development and validation are guided by rigorous, interdisciplinary collaboration.
Xue-Xia Mao, Huajun Yang, Shiwen Weng et al.· Frontiers in Neuroscience· 0 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduAug 27, 2026
A new machine-learning framework aims to improve the success rate of computational protein design while moving away from results that reproduce sequences found in nature.
