A scalable and efficient digital signal processing system for real time biological spike detection

The design of computing systems able to process data from electrophysiology experiments in real-time is required by biological applications where feedback stimulation of cells could depend on their current activities. From algorithm to hardware implementations, these closed loop systems are specific to cell types, culture size and analog acquisition system properties. Consequently, developing digital processing systems for biological experimentation is time consuming and could become a limiting factor for research in neuroscience. Model-based methodologies for hardware architecture generation is currently investigated in video and signal processing domains because they drastically reduce the development time of hardware prototypes and improve the design flexibility with low over-cost in comparison with handmade designs. In this article, we evaluate the interests and the drawbacks of these model-based methodologies for electrophysiological applications and compare them with previous handmade design.