Preclinical rodent models of neurological disorders are imperative to study the efficiency of novel therapeutics.
The proposed research work is set in the framework of Epilepsy in a collaboration of the Behavioural Neuroscience Laboratory (https://behavioural-neuroscience.com/) and the Molecular Neurophysiology and Epilepsy group (https://portal.research.lu.se/en/organisations/molecular-neurophysiology...) .
Currently, clinical and preclinical trials use expert raters to assess the status and the severity of epileptic seizures. This type of rating is very time-consuming, subject to human bias and a relatively crude form of assessment. An automated integration of analysing movements will drastically increase data reliability and provide a more fine-grained analysis of behavioural states. Recent developments of the last years have increased the ability of AI/ML to process a large number of images in Convolutional Neural Networks. Using markerless pose estimation based on transfer learning we can reach human accuracy using neural networks in the behavioural phenotyping.
In our labratory we have recentlty constructed a testing platform which allows us to obtain pose-estimation data from animals. The main part of the project would be to train a new CNN to classify different disease states. Hence, relevant knowledge in ML and Python would be required for this internship.