RADIOMICS IN PET/CT: NUMBERS CAN TELL WHAT CANNOT BE SEEN!

The adoption of precision medicine concept has speeded up the progression of high throughput quantitative metrics with substantial methodological advancement in technologies interrogating biological systems and paving the way for “omics” development. “Radiomics” is one of the “omics” harvesters. As by extraction of multiple features from medical images using methods from bioinformatics, additional information predicting tumor biology, can be obtained. It was first proposed by Philippe Lambin in 2012 and since then considered an evolving field in medical imaging in oncology. Numerous studies were carried out using various PET tracers for several tumors to investigate different radiomics signatures as the shape, intensity, and textural features aiming to providing precise risk stratification by integrating imaging features into prediction models of therapy outcome. Radiomics research has opened up a broad horizon for researchers to study new dimensions of medical imaging that appears to have an impact on clinical decision-making from an entirely new perspective. Hence Joint EANM/SNMMI has published a guideline on best practices for robust radiomics analyses. Recent years have been marked by rapid developments in the field of artificial intelligence, and radiomics has become an important sub-field and a very quickly evolving field of research as machine learning algorithms have greatly facilitated its existence in the real world. Meanwhile, further steps are still required to establish and standardize these techniques appropriately and which would pave the way for their introduction into routine clinical practice.