Träuble, J. | Search bioRxiv

bioRxiv2025

Enhancing Brain Age Prediction and Neurodegeneration Detection with Contrastive Learning on Regional Biomechanical Properties

The aging process affects brain structure and function, yet its biomechanical properties remain underexplored. Magnetic Resonance Elastography (MRE) provides a unique perspective by mapping brain tissue stiffness and damping ratio, observables that correlate with age and disease. Using a self-supervised contrastive regression framework, we demonstrate that MRE surpasses conventional structural magnetic resonance imaging (MRI) in sensitivity. Specifically, stiffness captures Alzheimers disease (AD), while damping ratio detects subtle changes associated with mild cognitive impairment (MCI). Our regional analysis identifies deep brain structures, particularly the caudate and thalamus, as key biomarkers of aging. The greater age sensitivity of MRE translates to superior differentiation of AD and MCI from healthy individuals, pinpointing regions where significant biomechanical alterations occur, notably the thalamus in AD and hippocampus in MCI. Furthermore, our results reveal biomechanical alterations in cognitively healthy individuals whose aging profiles closely resemble patients with MCI and AD. These findings highlight MREs potential as a biomarker for early neurodegenerative changes, aiding dementia risk detection and early intervention.

neuroscience↗

bioRxiv2024

Machine learning-based spike sorting reveals how subneuronal concentrations of monomeric Tau cause a loss in excitatory postsynaptic currents in hippocampal neurons

Extracellular recordings of neuronal activity constitute a powerful tool for investigating the intricate dynamics of neural networks and the activity of individual neurons. Microelectrode arrays (MEAs) allow for recordings with a high electrode count, ranging from 10s to 1000s, generating extensive datasets of neuronal information. Furthermore, MEAs capture extracellular field potentials from cultured cells, resulting in highly complex neuronal signals that necessitate precise spike sorting for meaningful data extraction. Nevertheless, conventional spike sorting methods face limitations in recognising diverse spike shapes, thereby constraining the full utilisation of the rich dataset acquired from MEA recordings. To overcome these limitations, we have developed a machine learning algorithm, named PseudoSort, which employs advanced self-supervised learning techniques, a distinctive density-based pseudo-labelling strategy, and an iterative fine-tuning process to enhance spike sorting accuracy. Through extensive benchmarking on large-scale simulated datasets, we demonstrate the superior performance of PseudoSort compared to recently developed machine learning-based (ML) spike sorting algorithms. We showcase the practical application of PseudoSort by utilising MEA recordings from hippocampal neurons exposed to subneuronal concentrations of monomeric Tau, a protein associated with Alzheimers disease (AD). Our results, validated against patch clamp experiments, unveil that monomeric Tau at subneuronal concentrations induces stimulation-dependent disruptions in both local and global activity of hippocampal neurons. Remarkably, patch clamp electrophysiology highlights the effect of combined Tau and neuronal stimulation treatment on excitatory postsynaptic currents, whereas PseudoSort excels in identifying neuronal clusters that exhibit diminished firing capacity following Tau treatment alone, i.e., in the absence of stimulation. This comprehensive approach validates the prowess of PseudoSort and unravels the intricate effects of Tau on neuronal activity, particularly in the context of AD.

neuroscience↗