Optogenetics in Silicon: A Neural Processor for Predicting Optically Active Neural Networks.

Area of Science:

• Neuroscience
• Computational Biology
• Electrical Engineering

Background:

• Opto-neural behavior simulation requires efficient computational models.
• Existing hardware often lacks the complexity to model ion channel dynamics accurately.

Purpose of the Study:

• To present a reconfigurable neural processor for real-time simulation and prediction of opto-neural behavior.
• To integrate detailed neuron and Channelrhodopsin-2 (ChR2) models into efficient hardware.

Main Methods:

• Developed a Field Programmable Gated Array (FPGA) architecture with a custom data-path, data management system, and memory-based router.
• Integrated a Hodgkin-Huxley CA3 neuron model with a four-state ChR2 model, including short and long-term calcium and light-dependent ion channels.
• Achieved computational efficiency with processing times of 0.03 ms/neuron and 9.7 ms/network (500 neurons) at 56.7 MHz.

Main Results:

• Successfully simulated opto-neural behavior in real-time using reconfigurable silicon hardware.
• Incorporated complex ion channel dynamics previously unavailable in hardware simulations.
• Demonstrated high computational efficiency for both single neurons and neural networks.

Conclusions:

• The developed processor enables efficient, biologically realistic simulation of opto-neural systems.
• Facilitates exploration of closed-loop processing and tuning of optogenetic circuitry.
• Represents a significant advancement in hardware for neuroscience research and optogenetics.