TUESDAY, March 26, 2019 (HealthDay News) -- For patients with lung adenocarcinoma, radiomic texture features extracted from within and around the nodule on baseline computed tomography (CT) scans can predict response to chemotherapy, according to a study published online March 20 in Radiology: Artificial Intelligence.
Mohammadhadi Khorrami, Ph.D., from the Case Western University School of Engineering in Cleveland, and colleagues retrospectively analyzed data from 125 patients with lung adenocarcinoma who had been treated with pemetrexed-based platinum doublet chemotherapy. The patients were divided randomly into two sets, with an equal number of responders and nonresponders in the training set. To predict response to chemotherapy, a machine learning classifier trained with radiomic texture features extracted from intratumoral and peritumoral regions of noncontrast enhanced CT images was used.
The researchers found that the mean maximum area under the receiver operating characteristic curve was 0.82 in the training set (53 patients) and 0.77 in the independent testing set (72 patients) for a combination of radiomic features in conjunction with a quadratic discriminant analysis classifier. There was a significant association for the radiomics signature with time to progression and overall survival (hazard ratios, 2.8 and 2.35, respectively). In terms of clinical usefulness, the radiomics signature had a higher overall net benefit than the clinicopathologic measurements in prediction of high-risk patients to receive treatment.
"Additional large-scale multisite validation needs to be performed before this radiomic model would be fit for clinical deployment," the authors write.
Two authors disclosed financial ties to the pharmaceutical and medical device industries.