ML in Oncology Research
Time-series Medical Image Analysis
Prof. Austin Duffy (MMUH), Dr. Caitriona McKendry (RCSI), Dr. Susan Maguire (Mater Private), Dr. Fiona Desmond (MMUH), Dr. John Duignan (SVUH) and Dr. Ronan Killeen (SVUH)
Nicholas McCarthy, Niamh Belton, Adil Dahlan, Misgina Hagos and Brendan Kelly
Cancer remains one of the leading causes of mortality globally despite the growing availability of high efficacy treatment options. The standard clinical approach to identify the best treatment is to iteratively trial treatments until a durable clinical benefit is achieved. This poses challenge, as clinicians race to find optimal treatments. A clear clinical need exists for predictive tools which can provide insight into the most likely treatments to provide durable clinical benefit.
Machine Learning and AI have the potential to revolutionise the clinical approach to selecting appropriate treatments by providing more accurate, early predictions of successful treatment for a given patient. We are leveraging both radiomic and deep learning approaches on multiple timepoint scans to deliver higher accuracy, early predictions and metrics which quantify response to treatment on a continuous basis.