A Machine Learning Model Based on Multi-Phase Contrast-enhanced CT for the Preoperative Prediction of the...

Area of Science:

• Radiology
• Oncology
• Artificial Intelligence

Background:

• Muscle infiltration is a key prognostic indicator for bladder cancer.
• Accurate identification of muscle invasion is crucial for treatment planning.
• Machine learning offers potential for precise image-based assessment of muscle infiltration.

Purpose of the Study:

• To develop and validate a machine learning model for preoperative evaluation of bladder cancer muscle invasiveness.
• Utilize multi-phase contrast-enhanced CT (MCECT) imaging for model development.
• Enhance diagnostic accuracy for muscle-invasive bladder cancer.

Main Methods:

• Retrospective analysis of MCECT scans from bladder cancer patients.
• Radiomics feature extraction and signature development using statistical methods and LASSO regression.
• Establishment and evaluation of machine learning classifiers (including logistic regression) using ROC curves, calibration, and decision curve analysis.

Main Results:

• The logistic regression model demonstrated strong predictive performance.
• Achieved an AUC of 0.89 in the training group, 0.80 in the test group, and 0.87 in the external testing group.
• Reported high diagnostic accuracy, precision, sensitivity, specificity, and F1-scores in testing cohorts.

Conclusions:

• The developed machine learning model accurately predicts muscle infiltration status in bladder cancer preoperatively.
• This AI-driven approach can improve the non-invasive assessment of bladder cancer staging.
• Potential to guide surgical and treatment strategies for bladder cancer patients.