We analyse MRI, PET and spectroscopic data to segment, classify and grade brain tumours, and to read radiogenomic and molecular signatures non-invasively, including the spectrobiopsy concept we pioneered in Australia.

Computational Neuroimaging is where the lab turns clinical scans into quantitative, decision-ready information. We develop and validate methods that analyse MRI, PET and spectroscopic data to segment, classify and grade brain tumours, and increasingly to read molecular and genetic signatures directly from imaging, non-invasively, before a scalpel is ever involved. A central thread of our work is radiomics and radiogenomics: extracting high-dimensional quantitative features from routine scans and linking them to underlying tumour biology, so that imaging can predict molecular subtype, aggressiveness and likely treatment response. Building on this, the lab pioneered the spectrobiopsy concept in Australia, the idea that spectroscopic imaging can act as a virtual biopsy, characterising tissue at the molecular level without surgical sampling. Underpinning all of this is a commitment to methods that are robust, reproducible and clinically translatable. We build standardised acquisition and analysis pipelines, deep-learning models for segmentation and grading, and validation across multi-institutional cohorts, so that what works in a paper has a path to working at the bedside.