Radiomics is a novel approach for making the most of medical images. Assuming that the content of medical images is currently underused, this approach consists in extracting a large number of parameters from the images and investigating their relationships with other characteristics of the disease (omics data, prognosis, aso). In PET using Fluorodeoxyglucose (FDG), several groups have proposed to extract textural indices from tumor hypermetabolic regions in an attempt to identify parameters that could predict the response to therapy or patient survival. Yet, in spite of encouraging results, the biological meaning of textural indices in PET has never been studied.

In an article in press in the Journal of Nuclear Medicine, researchers from IMIV thoroughly investigated the potential and limitations of textural indices derived from PET images using mice bearing orthopically implanted mammary tumors. They found that texture indices calculated in vivo from FDG PET images were highly correlated with the same indices measured on ex vivo autoradiography images that have a much higher spatial resolution. In addition, textural indices were sensitive to the tumor cell arrangement and density as seen on the corresponding histological slides. 

These results obtained by Fanny Orlhac under the supervision of Irène Buvat are the first evidence of the biological relevance of textural indices calculated from PET images to characterize tumor metabolic heterogeneity, paving the way towards their future use in the clinics.