An Automated Squint Method for Time-syncing Behavior and Brain Dynamics in Mouse Pain Studies.

Area of Science:

• Neuroscience
• Computational Biology
• Animal Models

Background:

• Quantifying spontaneous pain, particularly head pain in conditions like migraine, presents challenges due to real-time tracking difficulties and human bias.
• Eye squint has emerged as a promising, continuously measurable metric for predicting pain states in preclinical assays.

Purpose of the Study:

• To develop and validate a protocol for automating and quantifying eye squint in mice using DeepLabCut (DLC).
• To enable unbiased, real-time measurement of eye squint for comparison with neurophysiological data.
• To assess AI training parameters for accurate discrimination of squint and non-squint states.

Main Methods:

• Utilized DeepLabCut (DLC), a markerless deep learning tool, for automated eye tracking.
• Quantified eye squint by measuring the Euclidean distance between eyelids in restrained mice.
• Assessed the impact of AI training parameters on model performance.
• Validated the protocol in a CGRP-induced migraine-like mouse model.

Main Results:

• Successfully developed a protocol for automated, unbiased quantification of eye squint using DLC.
• Demonstrated reliable tracking and differentiation of squint and non-squint periods at sub-second resolution.
• Identified key AI training parameters crucial for successful model performance.

Conclusions:

• The DLC-based protocol offers an objective and reliable method for real-time pain assessment in mice.
• This approach facilitates the integration of behavioral pain metrics with mechanistic neurophysiological measures.
• The automated quantification of eye squint holds significant potential for advancing migraine research and drug development.