Speaker
Description
We present a hybrid multiscale model that establishes a mechanistic link between tumour microenvironment dynamics and FDG-PET radiomic signatures. The key innovation is the integration of multiscale tumour modelling with synthetic PET-like image generation, enabling PET radiomic signatures to be interpreted in terms of underlying biological processes. The model integrates intracellular signalling, cell-based tumour dynamics, and tissue-scale transport to investigate how cellular heterogeneity and microvascular organization shape SUV-based, morphological, and texture-derived features across tumour progression. We compare homogeneous and heterogeneous tumour scenarios and show that distinct microenvironmental conditions give rise to measurable differences in PET radiomic patterns. In vascularized tumours, chemotherapy induces reductions in compactness, mean uptake, and texture contrast, consistent with decreased tumour burden and altered spatial heterogeneity. Agreement with published vascular sprout dynamics, together with consistency between synthetic and clinical lung PET/CT radiomic ranges, supports the biological plausibility of the framework. These findings highlight the potential of mechanistic modelling to provide biologically grounded interpretation of PET radiomics and improve quantitative assessment of therapy response.
