Classification of pseudocalcium visual responses from mouse retinal ganglion cells

Area of Science:

• Neuroscience
• Computational Biology
• Biomedical Engineering

Background:

• A catalog of 32 mouse retinal ganglion cell (RGC) visual response patterns exists, based on calcium imaging.
• Acquiring sufficient calcium imaging data for bionic vision research is challenging.
• Extracellular spike train recordings are more feasible but difficult to directly compare to calcium-based RGC classifications.

Purpose of the Study:

• To convert extracellular spike train recordings into pseudocalcium signals.
• To enable direct comparison of spike train data with existing calcium signal-based RGC catalogs.
• To facilitate bionic vision research by reconciling different neural recording methods.

Main Methods:

• Recorded spike trains from mouse RGCs using a microelectrode array (MEA).
• Applied visual stimuli including moving bars, contrast/frequency chirps, and color flashes.
• Conducted pseudocalcium conversion using an OGB-1 convolution kernel on spike train histograms.
• Utilized sparse principal components analysis for response feature extraction and RGC classification.

Main Results:

• Pseudocalcium signals were generated from spike trains.
• The converted signals partially mapped to the 32 predefined RGC groups.
• Some RGC groups remained unmatched, indicating limitations of the pseudocalcium approach for MEA data.

Conclusions:

• Adapting the pseudocalcium method for MEA spike train recordings was partially successful.
• New classification methods are needed to define RGC groups from MEA data for bionic vision.
• This study provides guidance on the limitations and potential of pseudocalcium signal conversion for future research.