Biology subjects

Rittman, T.

Publications and source records attributed to Rittman, T..

5 recordsLinked to original sources

bioRxiv2026

From group-level neurophysiological biomarkers to individual-level phenotyping to identify mechanisms of neurodegenerative disease

Precision medicine approaches require physiological effects to be informative when translated to the individual. Reliable group differences may have limited individual value when distributions overlap, mechanisms are heterogeneous, or patients differ in phenotype. A feature can differ between group means and still overlap heavily across individuals Moreover, combinations of physiological abnormalities may reveal subject-level structure that is not captured by any single marker at the group level. We evaluate the translation between group- and individual-level features derived from resting-state magnetoencephalography (MEG) using behavioural-variant frontotemporal dementia (bvFTD) and progressive supranuclear palsy (PSP) as demonstrator conditions. These illnesses have distinct molecular pathology but share the pressing need for therapeutic development and have partially overlapping clinical and physiological features. We combine clinical phenotype, power spectral density (PSD), and latent network mechanisms inferred by dynamic causal modelling (DCM) to compare groups and to access their individuals mapping translations. We used classification as an exploratory probe of how group-defined physiological features from 74 healthy controls, 45 participants with bvFTD and 55 participants with PSP support individual assignment. Clinical scores showed strong individual-level translation, consistent with clinical diagnostic definitions (e.g, controls-vs-PSP accuracy 0.94; bvFTD-vs-PSP accuracy 0.83). MEG spectra feature classified individuals weakly (accuracy ~0.6), despite group level re-organization (e.g., reduced beta-band and increased fronto-insular alpha-phase/beta-amplitude coupling in bvFTD). Inferred biophysical parameters gave intermediate performance (e.g. PSP-vs-control accuracy 0.70; bvFTD-vs-PSP accuracy 0.74) while retaining mechanistic interpretability. Using this approach, experimental medicine studies of neurological disease can formally trade between mechanistic insight and classification performance, in using biomarkers that can account for individual heterogeneity within and between diagnostic groups.

bioengineering↗

bioRxiv2026

A Bayesian approach to volumetric analysis in rare neurodegenerative disorders

Identifying regional brain atrophy is critical for understanding neurodegenerative disorders. Traditionally, group comparisons rely on frequentist statistics, utilising $p$-values and point estimates that offer limited characterization of uncertainty. Bayesian methods provide an alternative by estimating effect sizes and credible intervals while incorporating prior knowledge and uncertainty. This study evaluated the feasibility of Bayesian hierarchical modelling to assess regional brain atrophy in Progressive Supranuclear Palsy (PSP) and Corticobasal Syndrome (CBS) compared to healthy controls. Participants meeting diagnostic criteria for CBS or PSP, together with healthy controls, were recruited from the Cambridge Centre for Parkinson-plus. Grey matter volume, cortical thickness, and subcortical volumes were derived using FreeSurfer for 210 cortical and 36 subcortical regions defined by the Human Brainnetome Atlas. Participants with PSP were classified into Richardson syndrome (PSP-RS), corticobasal syndrome variant (PSP-CBS), frontal variant (PSP-F), and other presentations (PSP-Other). Bayesian hierarchical ANOVA models with appropriate priors were used to estimate group differences. Regions of practical equivalence (ROPE) were calculated to assess evidence of meaningful group effects. A total of 428 people were included (113 PSP-RS, 30 PSP-CBS, 23 PSP-Other, 19 PSP-F, 81 CBS, and 162 healthy controls). For grey matter volume, there was evidence of whole brain volume loss compared to controls for all disease groups except PSP-Other, which showed evidence of no difference. For cortical thickness, there was evidence of global cortical thinning in CBS compared to controls, but for no other group. For subcortical volume all groups showed evidence of volume loss compared to controls. Patterns of regional atrophy were identified in keeping with previous literature, with the largest effect observed in the left pre-motor thalamus (30\% reduction in CBS). There was greater loss of subcortical volume in PSP-RS compared to other PSP subgroups. Bayesian hierarchical modelling provides a feasible framework for quantifying regional brain atrophy, with interpretable effect size estimates and uncertainty measures. This approach complements conventional statistical methods, particularly in the assessment of rare neurodegenerative and neuropsychiatric disorders.

neuroscience↗

bioRxiv2025

Uncertainty-Aware Tau Detection in Progressive Supranuclear Palsy Using Object Detection Models

Abnormal tau accumulation is a hallmark of neurodegenerative tauopathies such as Progressive Supranuclear Palsy (PSP). Traditional post-mortem assessments rely on manual lesion annotation, which is time-consuming and subjective. Existing machine learning methods typically involve multi-stage, feature-based pipelines, resulting in limited scalability and reliance on handcrafted features. This work evaluates deep learning-based object detection (OD) as a unified, end-to-end alternative for automating tau lesion identification and classification. We fine-tuned YOLO and Faster R-CNN models on a dataset of 29,852 tau lesions from 16 PSP brain slides, achieving mAP@50 of 0.702. To address the limitations of conventional OD models, which assume overconfident spatial predictions, we integrated Monte Carlo Dropout into both architectures to estimate spatial and label uncertainty. We evaluated detection uncertainty using Probability-based Detection Quality. Our results demonstrate that incorporating uncertainty offers a more comprehensive evaluation of detection reliability, complementing traditional metrics like mAP. For clinical application, we integrate a non-stochastic YOLO model into QuPath for whole-slide inference. Downstream clinical validation was conducted, including SHAP-based saliency analysis and correlation between model-predicted tau densities and PSP Rating Scale scores, a clinical disease severity measure. The alignment of detection outputs with pathological features and clinical severity underscores the clinical utility of OD models. Overall, this work highlights the complementary strengths of deep object detectors and uncertainty quantification, offering a path toward scalable, interpretable, and clinically meaningful tau pathology analysis. Code is available at https://github.com/Pidongg/dl_histopathology.

bioinformatics↗

bioRxiv2020

Amplification, not spreading limits rate of tau aggregate accumulation in Alzheimer's disease

Both the replication of protein aggregates and their spreading throughout the brain are implicated in the progression of Alzheimers disease (AD). However, the rates of these processes are unknown and the identity of the rate-determining process in humans has therefore remained elusive. By bringing together chemical kinetics with measurements of tau seeds and aggregates across brain regions, we are able to quantify their replication rate in human brains. Remarkably, we obtain comparable rates in several different datasets, with 5 different methods of tau quantification, from seed amplification assays in vitro to tau PET studies in living patients. Our results suggest that the overall rate of accumulation of tau in neocortical regions is limited not by spreading between brain regions but by local replication, which doubles the number of seeds every ~5 years. Thus, we propose that limiting local replication constitutes the most promising strategy to control tau accumulation during AD.

biophysics↗

bioRxiv2018

In vivo assay of cortical microcircuitry in frontotemporal dementia: a platform for experimental medicine studies

The analysis of neural circuits can provide critical insights into the mechanisms of neurodegeneration and dementias, and offer potential quantitative biological tools to assess novel therapeutics. Here we use behavioural variant frontotemporal dementia (bvFTD) as a model disease. We demonstrate that inversion of canonical microcircuit models to non-invasive human magnetoecphalography can identify the regional- and laminar-specificity of bvFTD pathophysiology, and their parameters can accurately differentiate patients from matched healthy controls. Using such models, we show that changes in local coupling in frontotemporal dementia underlie the failure to adequately establish sensory predictions, leading to altered prediction error responses in a cortical information-processing hierarchy. Using machine learning, this model-based approach provided greater case-control classification accuracy than conventional evoked cortical responses. We suggest that this approach provides an in vivo platform for testing mechanistic hypotheses about disease progression and pharmacotherapeutics.

neuroscience↗