Self-Supervised Foundation Model for Calcium-imaging Population Dynamics

2026-04-03

AI Summary

Existing calcium imaging analysis methods are typically task-specific and lack transferability across diverse applications. This work proposes CalM, a self-supervised foundation model tailored for calcium signals, which constructs a shared discrete vocabulary via a high-fidelity quantizer and employs a dual-axis autoregressive Transformer to jointly model dependencies across both neuronal and temporal dimensions. CalM introduces the first self-supervised pretraining paradigm specifically designed for calcium imaging data, achieving substantial performance gains over task-specific baselines on large-scale, multi-animal datasets. The model supports a range of downstream tasks—including prediction and neural decoding—and enables interpretable insights into functional neural structures through linear representation analysis.

Abstract

Recent work suggests that large-scale, multi-animal modeling can significantly improve neural recording analysis. However, for functional calcium traces, existing approaches remain task-specific, limiting transfer across common neuroscience objectives. To address this challenge, we propose \textbf{CalM}, a self-supervised neural foundation model trained solely on neuronal calcium traces and adaptable to multiple downstream tasks, including forecasting and decoding. Our key contribution is a pretraining framework, composed of a high-performance tokenizer mapping single-neuron traces into a shared discrete vocabulary, and a dual-axis autoregressive transformer modeling dependencies along both the neural and the temporal axis. We evaluate CalM on a large-scale, multi-animal, multi-session dataset. On the neural population dynamics forecasting task, CalM outperforms strong specialized baselines after pretraining. With a task-specific head, CalM further adapts to the behavior decoding task and achieves superior results compared with supervised decoding models. Moreover, linear analyses of CalM representations reveal interpretable functional structures beyond predictive accuracy. Taken together, we propose a novel and effective self-supervised pretraining paradigm for foundation models based on calcium traces, paving the way for scalable pretraining and broad applications in functional neural analysis. Code will be released soon.

Problem

• calcium imaging
• neural population dynamics
• foundation model
• self-supervised learning
• cross-task transfer

Innovation

• self-supervised learning
• foundation model
• calcium imaging
• dual-axis transformer
• neural population dynamics