Radiomic model approach for characterizing nodules promising
Tobias Peikert, M.D., from the Mayo Clinic in Rochester, Minnesota, and colleagues used the National Lung Screening Trial dataset to develop independent quantitative variables assessing radiologic nodule features using 726 indeterminate nodules (318 benign and 408 malignant). To enhance the prediction accuracy and interpretability of the multivariable model, analysis was performed using the least absolute shrinkage and selection operator (LASSO) method for variable selection and regularization.
The researchers found that LASSO multivariate modeling selected eight of the originally considered 57 quantitative radiologic features. These features included variables capturing location (vertical location); size (volume estimate); shape (flatness); density (texture analysis); and surface characteristics (surface complexity and estimates of surface curvature), all with P < 0.001. For the eight features, the optimism-corrected area under the curve was 0.939.
"Our novel radiomic LDCT-based approach for indeterminate screen-detected nodule characterization appears extremely promising however independent external validation is needed," the authors write.
Some study authors are co-inventors of an LDCT-based radiomic classifier for lung adenocarcinomas, distinct from the present work, licensed to Imbio Inc.
Copyright © 2018 HealthDay. All rights reserved.