Hypoxia is an important prognostic indicator in most solid tumours. We present here automated, data-driven methods, using principal component analysis (PCA) and Gaussian mixture modelling (GMM), that consistently locate functionally distinct sub-regions in preclinical tumours, some of which are postulated to be relevant to hypoxia. Methods are based on dynamic contrast-enhanced (DCE)-MRI (reflecting perfusion) and oxygen-enhanced (OE)-MRI (reflecting oxygen delivery). We demonstrate the utility and stability of our methods through a combination of evaluation metrics, which may be incorporated in similar studies elsewhere.
This abstract and the presentation materials are available to members only; a login is required.