Challenges and Promises of Radiomics for Rectal Cancer

Purpose of Review: This literature review aims to gather the relevant works published on the topic of Radiomics in Rectal Cancer. Research on this topic has focused on finding predictors of rectal cancer staging and chemoradiation treatment response from medical images. The methods presented may, in principle, aid clinicians with the appropriate treatment planning options. Finding appropriate automatic tools to help in this task is very important, since rectal cancer has been considered one of the most challenging oncological pathologies in recent years. Recent Findings: Radiomics is a class of methods based on the extraction of mineable, high-dimensional data/features from the routine, standard-of-care medical imaging. This data is then fed to machine learning algorithms, with the goal of automatically obtaining predictions regarding disease stage and therapeutic response. Summary: The literature reviewed suggests that Radiomics will continue to be a part of the body of research in oncology in the upcoming years. However, and excluding very few studies, proper validation on the performance of the methods (mainly with external datasets) is still one of the main limitations of the field, which strongly limits their clinical applicability. Progress will only occur if the community opens itself to collaborate with different groups, as data availability and limited shareability continues to be the barrier for its development. Nowadays, Radiomics is used for nearly every type of cancer. In particular, for rectal cancer, the need for predicting treatment response will continue to demand and boost research in this field.