Deep-learning based tracking of behaviour in preclinical models for mental illness
The Synapse Biology & Cognition Laboratory is focused on understanding the critical role synaptic genes and proteins play in shaping excitatory/inhibitory wiring and connectivity in the brain, that enables complex cognition and higher order processing in the healthy brain, and in mental conditions where these processes go awry.
Human genetic studies continue to increasingly highlight that disruption of postsynaptic genes is a hub for a range of mental health disorders, namely neurodevelopmental and neuropsychiatric disorders. These include schizophrenia, anxiety and mood disorders (Depression, Bipolar) and Autism Spectrum Disorders, that share overlapping symptom domains. While the importance of postsynaptic proteins in synaptic function and plasticity are strongly appreciated, we know much less about the impact of postsynaptic gene mutations in regulating distinct components of cognition and higher order processing.
Modelling the complex cognitive processes routinely assessed in the clinical setting has been challenging in animal models. Bridging the gap between preclinical and clinical cognitive testing, the touchscreen methodology that A/Prof Nithianantharajah was involved in early during its development, application and commercialisation at the University of Cambridge, provides an innovative tool for dissecting higher cognitive functions in rodents that is highly analogous to cognitive assessment of clinical populations.
Identifying the disrupted neural mechanisms that underlie cognitive symptoms in mental illnesses like schizophrenia and mood disorders remains a challenge for the development of novel effective treatments. The ability to measure and control behaviour in preclinical models, using automated behavioural systems while recording real-time neural activity provides advanced experimental approaches to tackle this challenge. Combining these approaches with novel tools for pose estimation with deep learning now allows training of deep neural networks to accurately quantify a range of complex behavioural measures.
This project will use deep learning tools (such as DeepLabCut) to analyse disrupted decision making and learning behaviour in genetic mouse models for complex mental illnesses.