Medicine has undoubtedly come a long way, yet a cure for cancer remains elusive. But new research suggests combining medical imaging data with cancer biology information may help.
That’s according to experts out of UCLA, who unveiled their "radiogenomic" neural network framework May 8 in the Journal of Medical Imaging. The model proved capable of learning about significant associations between subsets of genes and corresponding types of lung cancer.
Although other models have produced similar associations, this neural network achieved more, which may prove significant for radiology, the authors noted.
“We hope this approach increases the radiologist's confidence in assessing the type of lung cancer seen on a CT scan,” William Hsu, associate professor of radiological sciences at the Los Angeles-based university, explained on Monday. “This information would be highly beneficial in informing individualized treatment planning."
Building off prior work which utilized a small brain tumor dataset, Hsu and colleagues trained their neural network on a set of 262 patients consisting of more than 21,000 gene expressions. In total, the platform predicted 101 features.
They further tested its predictive ability on an independent set of 89 patients and used a method called “gene masking” to understand how the neural network learned to make its associations between gene subsets and lung cancer.
Predicting each imaging feature of cancer, it turns out, came down to the individual gene expression profile. Their technique is also more generalizable than past attempts and applicable to other patient populations.
"While radiogenomic associations have previously been shown to accurately risk stratify patients, we are excited by the prospect that our model can better identify and understand the significance of these associations,” Hsu said in a statement.
Matt joined Chicago’s TriMed team in 2018 covering all areas of health imaging after two years reporting on the hospital field. He holds a bachelor’s in English from UIC, and enjoys a good cup of coffee and an interesting documentary.