In kidney cancer, Stage 1 is an important threshold for the decision of organ preservation surgery versus chemotherapy and organ removal for higher stages. The purpose of this study is to compare the classification accuracy of two different Inception V3 deep learning neural networks, trained on cropped computer tomography (CT) images of either whole kidney containing cancer (DLNN- WK) or only the kidney cancer (DLNN-OC). The National Cancer Institute TCIA database provided anonymized 3D CT scans and clinical data from 227 patients for the training and testing of the DLNN-WK and DLNN-OC. The dataset was split into 48% training, 10% validation, and 42% testing sets. The area under the ROC curve (AUC) for the DLNN-WK was 0.96 for training, 0.88 for validation and 0.87 for test sets. The AUC for the DLNN-OC was of 0.97 for training, 0.91 for validation, and 0.90 for test sets. Both AI systems show promise for potentially assisting physicians in kidney cancer staging.
Journal: TechConnect Briefs
Volume: TechConnect Briefs 2021
Published: October 18, 2021
Pages: 161 - 164
Industry sector: Medical & Biotech
Topics: Biomaterials, Diagnostics & Bioimaging