CHICAGO—The disciplines of radiomics and genomics are being spliced for specialized tumor mapping that moves toward a less invasive, higher-tech supplement, and perhaps one day an alternative to biopsy, according to a molecular imaging symposium presented Dec. 2 at the Radiological Society of North America’s (RSNA) 99th annual meeting.
The newest radiogenomics applications correlate not just information about heterogeneity, but molecular imaging phenotypes based on a broad spectrum of genetic expressions clustered into “metagenes” associated with particular cancers. Certain metagenes are known to be associated with specific features. These features run the gamut of morphology, shape and texture. The goal is a fusion of complex imaging and genetic data that provides not just comprehensive staging of disease, but the possibility to predict patient outcomes. All told, it is a fragile endeavor, according to Olivier Gevaert, PhD, assistant professor of medicine and biomedical informatics research at Stanford University in Stanford, Calif.
“If you don’t get it right, you’re going to destroy clinical data,” he said.
For instance, features such as sharpness and axis are related to specific outcomes in patients. In the case of non-small cell lung cancer, a model created from 180 different features includes 26 individual tumor traits that are correlated with patient survival. This, in turn, provides a means of multivariate survival modeling, but it is still a work in progress. Molecular imaging is not yet at a stage where data gleaned from image phenotyping can replace the genetic data obtained from biopsy, but working in tandem, radiomics and genomics stand to provide a more complete clinical picture, according to Gevaert.