Machine Learning Classification of Patients Undergoing Abdominal CT Scans: Factors Affecting Imaging Quality and Radiation Exposure

The increasing use of computed tomography scans in medical diagnostics necessitates a delicate balance between achieving high imaging quality and minimising radiation exposure. This study investigates the application of machine learning algorithms to classify different CT scan types and elucidate the factors influencing imaging quality and radiation exposure in abdominal CT scans. The GE Optima 660 CT scanner collected a dataset encompassing patient demographics, exposure parameters, and dose data. ML algorithms were employed for classification, including kNearest Neighbors, Support Vector Machine, Logistic Regression, Random Forest, and Neural Network. The analysis involved a three-stage approach: initial classification without a specified target variable, a reduced training dataset to assess algorithm robustness, and variable importance determination using Logistic Regression. The ML algorithms demonstrated notable performance in classifying CT scan types, with SVM, Random Forest, and Neural Networks achieving perfect scores. Key variables such as Low kilovoltage (kV), Tube current modulation (mA), and contrast material emerged as decisive in the classification process. Notably, their significance remained consistent across both contrast and non-contrast scans. The study's outcomes underscore the potential of machine learning in balancing image quality and radiation exposure, aligning with the healthcare goal of patient safety. The consistency in variable importance across different imaging scenarios highlights the reliability of the machine learning model. The implications for healthcare efficiency are substantial, enabling informed decisions for protocol optimisation based on patient-specific factors. These findings align with broader discussions on AI's role in reducing CT radiation dose, emphasising the significance of machine learning in achieving diagnostic precision with lower radiation exposure.