arXiv subjects

Chetan Gohil

Publications and source records attributed to Chetan Gohil.

5 recordsLinked to original sources

arXiv2026

A Systematic Evaluation of Sample-Level Tokenization Strategies for MEG Foundation Models

Recent success in natural language processing has motivated growing interest in large-scale foundation models for neuroimaging data. Such models often require discretization of continuous neural time series data, a process referred to as 'tokenization'. However, the impact of different tokenization strategies for neural data is currently poorly understood. In this work, we present a systematic evaluation of sample-level tokenization strategies for transformer-based large neuroimaging models (LNMs) applied to magnetoencephalography (MEG) data. We compare learnable and non-learnable tokenizers by examining their signal reconstruction fidelity and their impact on subsequent foundation modeling performance (token prediction, biological plausibility of generated data, preservation of subject-specific information, and performance on downstream tasks). For the learnable tokenizer, we introduce a novel approach based on an autoencoder. Experiments were conducted on three publicly available MEG datasets spanning different acquisition sites, scanners, and experimental paradigms. Our results show that both learnable and non-learnable discretization schemes achieve high reconstruction accuracy and broadly comparable performance across most evaluation criteria, suggesting that simple fixed sample-level tokenization strategies can be used in the development of neural foundation models. The code is available at https://github.com/OHBA-analysis/Cho2026_Tokenizer.

cs.LG↗

arXiv2026

Estimating measures of information processing during cognitive tasks using functional magnetic resonance imaging

Cognition is increasingly framed in terms of information processing, yet most fMRI analyses focus on activation or functional connectivity rather than quantifying how information is stored and transferred. To remedy this problem, we propose a framework for estimating measures of information processing: active information storage (AIS), transfer entropy (TE), and net synergy from task-based fMRI. AIS measures information maintained within a region, TE captures directed information flow, and net synergy contrasts higher-order synergistic to redundant interactions. Crucially, to enable this framework we utilised a recently developed approach for calculating information-theoretic measures: the cross mutual information. This approach combines resting-state and task data to address the challenges of limited sample size, non-stationarity and context in task-based fMRI. We applied this framework to the working memory (N-back) task from the Human Connectome Project (470 participants). Results show that AIS increases in fronto-parietal regions with working memory load, TE reveals enhanced directed information flows across control pathways, and net synergy indicates a global shift to redundancy. This work establishes a novel methodology for quantifying information processing in task-based fMRI.

q-bio.NC↗

arXiv2025

MEG-GPT: A transformer-based foundation model for magnetoencephalography data

Modelling the complex spatiotemporal patterns of large-scale brain dynamics is crucial for neuroscience, but traditional methods fail to capture the rich structure in modalities such as magnetoencephalography (MEG). Recent advances in deep learning have enabled significant progress in other domains, such as language and vision, by using foundation models at scale. Here, we introduce MEG-GPT, a transformer based foundation model that uses time-attention and next time-point prediction. To facilitate this, we also introduce a novel data-driven tokeniser for continuous MEG data, which preserves the high temporal resolution of continuous MEG signals without lossy transformations. We trained MEG-GPT on tokenised brain region time-courses extracted from a large-scale MEG dataset (N=612, eyes-closed rest, Cam-CAN data), and show that the learnt model can generate data with realistic spatio-spectral properties, including transient events and population variability. Critically, it performs well in downstream decoding tasks, improving downstream supervised prediction task, showing improved zero-shot generalisation across sessions (improving accuracy from 0.54 to 0.59) and subjects (improving accuracy from 0.41 to 0.49) compared to a baseline methods. Furthermore, we show the model can be efficiently fine-tuned on a smaller labelled dataset to boost performance in cross-subject decoding scenarios. This work establishes a powerful foundation model for electrophysiological data, paving the way for applications in computational neuroscience and neural decoding.

cs.LG↗

arXiv2025

Cross Mutual Information

Mutual information (MI) is a useful information-theoretic measure to quantify the statistical dependence between two random variables: $X$ and $Y$. Often, we are interested in understanding how the dependence between $X$ and $Y$ in one set of samples compares to another. Although the dependence between $X$ and $Y$ in each set of samples can be measured separately using MI, these estimates cannot be compared directly if they are based on samples from a non-stationary distribution. Here, we propose an alternative measure for characterising how the dependence between $X$ and $Y$ as defined by one set of samples is expressed in another, \textit{cross mutual information}. We present a comprehensive set of simulation studies sampling data with $X$-$Y$ dependencies to explore this measure. Finally, we discuss how this relates to measures of model fit in linear regression, and some future applications in neuroimaging data analysis.

cs.IT↗

arXiv2024

osl-ephys: A Python toolbox for the analysis of electrophysiology data

We describe OHBA Software Library for the analysis of electrophysiological data (osl-ephys). This toolbox builds on top of the widely used MNE-Python package and provides unique analysis tools for magneto-/electro-encephalography (M/EEG) sensor and source space analysis, which can be used modularly. In particular, it facilitates processing large amounts of data using batch parallel processing, with high standards for reproducibility through a config API and log keeping, and efficient quality assurance by producing HTML processing reports. It also provides new functionality for doing coregistration, source reconstruction and parcellation in volumetric space, allowing for an alternative pipeline that avoids the need for surface-based processing, e.g., through the use of Fieldtrip. Here, we introduce osl-ephys by presenting examples applied to a publicly available M/EEG data (the multimodal faces dataset). osl-ephys is open-source software distributed on the Apache License and available as a Python package through PyPi and GitHub.

q-bio.QM↗