SimSort: A Powerful Framework for Spike Sorting by Large-Scale Electrophysiology Simulation

2025-02-05

AI Summary

Spike sorting has long suffered from the lack of ground-truth labels in real neural recordings, hindering rigorous validation of heuristic approaches. To address this, we propose SimSort: a data-driven framework grounded in large-scale biophysical simulation. SimSort generates high-fidelity electrophysiological synthetic datasets, designs an end-to-end deep learning pretraining paradigm, and introduces a novel zero-shot transfer strategy—enabling spike sorting without any real-world annotations. Our method jointly models multi-channel waveforms, incorporates realistic neuronal dynamics, and leverages self-supervised pretraining. Evaluated across multiple real extracellular electrode datasets, SimSort consistently outperforms state-of-the-art methods in accuracy, robustness to noise and drift, and cross-platform generalizability. By circumventing reliance on ground truth, SimSort overcomes the fundamental evaluation bottleneck imposed by label scarcity, establishing a verifiable, scalable paradigm for experimental neuroscience.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Bio-inspired Learning
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Web Mining and Content Analysis: Web data generation and simulation
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Spike sorting is an essential process in neural recording, which identifies and separates electrical signals from individual neurons recorded by electrodes in the brain, enabling researchers to study how specific neurons communicate and process information. Although there exist a number of spike sorting methods which have contributed to significant neuroscientific breakthroughs, many are heuristically designed, making it challenging to verify their correctness due to the difficulty of obtaining ground truth labels from real-world neural recordings. In this work, we explore a data-driven, deep learning-based approach. We begin by creating a large-scale dataset through electrophysiology simulations using biologically realistic computational models. We then present extbf{SimSort}, a pretraining framework for spike sorting. Remarkably, when trained on our simulated dataset, SimSort demonstrates strong zero-shot generalization to real-world spike sorting tasks, significantly outperforming existing methods. Our findings underscore the potential of data-driven techniques to enhance the reliability and scalability of spike sorting in experimental neuroscience.

Problem

• Develops deep learning for spike sorting
• Creates dataset via electrophysiology simulations
• Enhances spike sorting reliability and scalability

Innovation

• Deep learning-based spike sorting
• Large-scale electrophysiology simulations
• Zero-shot generalization capability