Investigating the use of a convolutional neural network for the automated segmentation of paediatric brain tumours

Student work

Tara Cogan (2019-20)

This project pioneered the use of a cutting-edge deep-learning technique (called “segmentation”) to distinguish tumour tissue from healthy tissue in MRI scans of paediatric brain tumour patients. While segmentation is widely used in medical imaging, this AI method had never been previously applied in the paediatric context. The results confirmed the viability of this technique, demonstrating how deep-learning models can find direct application in the clinic to aid in the rapid diagnosis and therapy planning, reducing human error and increasing the pace at which paediatric brain tumour patients can be treated.