Interrogating the mouse visual system by automated analysis of voluntary behaviour

Project background

Retinal degeneration and irreversible photoreceptor diseases affect ~220 million people worldwide and are currently incurable. As part of the disease pathology, light sensitive neurons die leaving the retina unable to respond to light and the patient blind. Typically, potential therapies are measured for effectiveness in mouse strains with retinal degeneration. Mouse visual perception is difficult to assess and is currently measured using a combination of electrophysiological and behavioural approaches. However, electrical recordings and behavioural reflex tests are only indirectly related to visual perception. More relevant behavioural tests exist but these tests require extensive animal training, are known to cause distress and are limited to association with a single visual stimulus. A comprehensive assessment of visual perception requires responses to be measured for many different stimuli.

Why we funded it

This David Sainsbury Fellowship aims to reduce the number of mice required to evaluate perceptual vision by developing spontaneous behavioural tests. Dr Storchi plans to develop a reliable method to track multiple body points of the mouse allowing recognition of spontaneous movements in response to varying visual stimuli. 

One of the current issues with analysing visual perception is the limitation to the number of trials per animal. This can lead to high trial-to-trial variability despite a statistically significant locomotory response to a visual stimulus. Using previously published data Dr Storchi has analysed statistical power to estimate the potential his refined behavioural test has for animal reduction. The test would allow trial numbers per animal to be increased, improving trial-to-trial variability. This will increase statistical power leading to a reduction of mice required to generate statistical significance. Approximately 12 mice per experiment are currently needed to generate statistical significance and Dr Storchi estimates his new methodology will reduce this requirement by 75%.

Research methods

Animal behaviours have been described as compositions of “behavioural modules” with a sequence of modules encoding a complex behaviour such as walking or rearing. Multidimensional tracking has been used as a higher throughput method to provide a comprehensive analysis of spontaneous behaviours for lower species such as Drosophila fruit flies and zebrafish. Dr Storchi aims to extend these methods to the mouse allowing analysis of both the animal’s location and the shape of its body. Changepoint analysis can then be used to determine changes between behavioural modules, which Dr Storchi can use to identify responses to visual stimuli as an evaluation of perceptual vision.