A hardware system for real-time decoding of in vivo calcium imaging data

Area of Science:

• Neuroscience
• Bioengineering
• Computational Biology

Background:

• Epifluorescence miniature microscopes (miniscopes) are crucial for in vivo calcium imaging of neural population activity.
• Online imaging and real-time decoding are essential for closed-loop experiments, requiring high computational efficiency and low latency.
• Existing systems face challenges in achieving the speed and accuracy needed for immediate feedback during behavioral tasks.

Purpose of the Study:

• Introduce DeCalciOn, an open-source device for real-time in vivo calcium imaging and population decoding.
• To provide a hardware-compatible solution for miniscopes using the UCLA Data Acquisition (DAQ) interface.
• To enable high-throughput, low-latency neural population decoding for closed-loop neuroscience research.

Main Methods:

• DeCalciOn integrates hardware and software for online motion stabilization, neural enhancement, and calcium trace extraction.
• Utilizes field-programmable gate array (FPGA) hardware for accelerated real-time image processing.
• Employs contour-free methods for efficient calcium signal extraction from sensor images, supporting decoding of up to 1024 traces per frame.

Main Results:

• Achieved decoding latencies of less than 50 ms after photon arrival at the miniscope sensor.
• Successfully decoded rat position on a linear track from hippocampal CA1 calcium signals (n=12).
• Accurately classified rat behaviors during an instrumental task using orbitofrontal cortex calcium signals (n=2).

Conclusions:

• DeCalciOn offers an affordable, plug-and-play solution for real-time calcium imaging and neural decoding in behaving animals.
• The system demonstrates high decoding accuracy at short latencies, advancing closed-loop experimental capabilities.
• Innovations in hardware acceleration and efficient signal processing enable rapid analysis of neural population activity.