Event-based imaging of neuronal and vascular dynamics in vivo

Abstract

Simultaneously acquiring high spatiotemporal resolution and wide field-of-view remains a major challenge when imaging functional neuronal dynamics. Conventional frame-based capture discards most acquired data during analysis while limiting temporal resolution, indicating fundamentally inefficient acquisition. Event cameras offer an attractive alternative for functional neural imaging by asynchronously detecting local brightness changes with sub-millisecond precision. Their high temporal resolution, high dynamic range, and significantly lower power consumption and data rates make event cameras ideally suited for functional imaging. We present the initial application of event cameras for functional biological imaging, from capturing vascular dynamics in the mouse cortex to calcium imaging in cultured neurons and in vivo brain tissue. Event cameras effectively track rapid vascular and neuronal dynamics across these diverse settings. To reconstruct continuous signals from sparse, asynchronous event streams, we implemented a self-supervised framework that leverages the precise temporal information in event data. Our framework opens new possibilities for high-temporal resolution, data-efficient functional imaging using event-based sensors.