A total of 173 patients were included in the study, 43 of whom had MIBC. After extraction and selecting radiomic features, 1,616 robust features were retained using the machine learning model. An additional 128 deep learning features were obtained and combined with radiomics features to produce a deep learning radiomics signature (DLRS). The results showed that the DLRS-based model using the Multi-layer Perceptron (MLP) had the best performance for detecting muscle invasion (training set AUC: 0.973260 (95% CI 0.9488-0.9978) and test set AUC: 0.884298 (95% CI 0.7831-0.9855)) outperforming the other models. In the test cohort, the sensitivity, specificity, and accuracy of the MLP model were 0.91 (95% CI 0.551-0.873), 0.78 (95% CI 0.594-0.863), and 0.58 (95% CI 0.729-0.827).
Using these techniques for accurate staging of bladder cancer and detection of MIBC is critical for proper clinical management. One potential advantage of these AI-assisted imaging analysis tools is to reduce variability from subjective assessments and inter-operator variations. Machine learning models are also highly cost-effective, saving time and labor. However, in-depth studies using larger cohorts across various centers will be necessary for validating the generalizability and accuracy of these tools. Moreover, prospective studies will indicate whether the widespread use of radiomics, enhanced CT, and machine learning analysis is feasible and practical in the real-world setting. Finally, developing methods for improving the interpretability of AI-based models will be necessary for adoption by clinicians.
Written by: Bishoy M. Faltas, MD, Director of Bladder Cancer Research, Englander Institute for Precision Medicine, Weill Cornell Medicine
Reference:
- Chen W, Gong M, Zhou D, et al. CT-based deep learning radiomics signature for the preoperative prediction of the muscle-invasive status of bladder cancer. Front Oncol. 2022;12:1019749. Published 2022 Dec 5. doi:10.3389/fonc.2022.1019749