Identification of Retinal Ganglion Cell Firing Patterns Using Clustering Analysis Supplied with Failure Diagnosis

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual Neuroscience

Background:

• Understanding neuronal population coding in the retina is crucial for visual function.
• Microelectrode arrays record spiking dynamics of retinal ganglion cells (RGCs).
• Current spike sorting software often requires manual intervention for quality control.

Purpose of the Study:

• To develop an automated and error-resilient spike sorting platform for RGCs.
• To improve the accuracy of spike sorting without human intervention.
• To enable reliable analysis of retinal signals for visual neuroscience research.

Main Methods:

• Utilized an integrated analytical platform for spike train detection and clustering of RGCs from mouse retinal recordings.
• Employed parametric and principal component feature spaces for spike projection and clustering.
• Implemented iterative algorithm refinement through failure diagnosis and rerunning clustering for enhanced accuracy.

Main Results:

• Achieved improved spike sorting accuracy and error resilience using the automated platform.
• Parametric features demonstrated superior performance over principal components, especially in low signal-to-noise conditions.
• Successfully identified active RGCs and their light responsiveness during visual stimulation.

Conclusions:

• The developed automated spike sorting approach enhances the reliability of retinal signal processing.
• This method offers a robust alternative to manual spike sorting, reducing potential errors.
• The platform is suitable for analyzing physiological signaling and could be applied to retinal neurodegenerative diseases.