The aim of this project is to create AI tools to improve the evaluation and planning of epilepsy surgery. Epilepsy is drug-resistant in around 1/3 of patients. For those whose epilepsy is driven by a focal lesion, surgical resection can be curative, but seizure freedom is only achieved in around 70%. It is not fully understood what determines outcome, but studies have shown that clinical variables such as duration of epilepsy, the underlying histopathology and complete resection of the lesion are significant factors.
This PhD project will aim to:
1) Create open source AI tools for detecting and classifying epilepsy lesions in a large, curated cohort of MRI scans.
2) Extend these tools to group epilepsy lesions into their newly-defined molecular diagnoses
3) Integrate this information into AI models predicting post-surgical outcome.
