So it’s been lovely to be invited to talk about radiomics here, and PET, because, as we all know, PET images are definitely more than just pictures, they’re actually data. And the whole purpose of radiomics is to extract quantitative information from those images and to look at what quantitative information in those images we can use to build clinical prediction models. We’ve known for a long time that things like metabolic tumor volume, maximum uptake in the image, and also more recently, disease dissemination markers for example, the maximum distance between lesions in the body are prognostic...
So it’s been lovely to be invited to talk about radiomics here, and PET, because, as we all know, PET images are definitely more than just pictures, they’re actually data. And the whole purpose of radiomics is to extract quantitative information from those images and to look at what quantitative information in those images we can use to build clinical prediction models. We’ve known for a long time that things like metabolic tumor volume, maximum uptake in the image, and also more recently, disease dissemination markers for example, the maximum distance between lesions in the body are prognostic. And what’s starting to happen is that we’re starting to be able to incorporate these types of radiomic features with clinical features, and also more recently with some genomic features to build these clinical prediction models. In particular, we as a group collaborating with colleagues in the PETRA Consortium, which is funded by Dutch Cancer Charity published in JCO last year, our International Metabolic Prognostic Index, which is actually comprised of the age, the metabolic tumor volume and the stage of the patient. And what we showed was that that performed much better than the clinical risk score, the IPI that we typically use. And the other advantage about that is because it’s a continuous score, we can actually use it to apply risk prediction for individuals. And more importantly, in clinical trial design, we can use it to work out what particular progression-free survival threshold, for example, we might want to think about testing novel approaches for treatment intensification