911±¬ÁÏÍø

Abstract: Paediatric brain cancer (PBC) is the second most common form of childhood cancer, yet it carries the highest mortality rate. Alongside the mortality rate, many survivors experience intellectual and physical disabilities long into adulthood. A clear, ground-truth diagnosis based upon molecular information can only be obtained from a tumour sample, which in brain cancers often necessitates risky surgery.

Current best clinical practice for PBC is that on initial identification of a tumour, a preliminary diagnosis is delivered by the clinician based upon MRI findings and other available clinical factors such as patient age and tumour location. The difficulty is that these preliminary diagnoses (and prognosis) are heavily dependent on clinical judgement, with experiential variation between clinicians leaving room for misdiagnosis. This project then aims to develop AI-based models, particularly deep learning, to provide enhanced diagnostic/prognostic prediction at this pre-surgical stage to enhance initial treatment planning. Jordan Vihermaki will be presenting key findings from his recently completed systematic review and meta-analysis evaluating the performance of machine learning and deep learning models for PBC diagnostic and prognostic prediction.