Optimization of pediatric multidetector CT imaging parameters using a machine learning–based Monte Carlo simulation model

Abstract

This study utilized Monte Carlo (MC) simulations to optimize radiation doses in pediatric multidetector computed tomography (MDCT) head scans by analyzing key parameters like tube current (mA), tube voltage (kV), pitch, and slice thickness. The findings indicate that reducing tube current significantly lowers the Computed Tomography Dose Index (CTDIvol) and Dose Length Product (DLP), effectively minimizing patient radiation exposure. Higher pitch values (0.7–0.9) further reduced radiation by decreasing beam overlap, while using a thinner slice thickness (0.6 mm) improved dose efficiency. A comparison highlighted the effectiveness of optimization: simulated parameters kVp 100, mAs 81, pitch 0.98 yielded a CTDIvol of 2.6 mGy and DLP of 131.3 mGy.cm, a substantial reduction compared to conventional clinical settings (kVp 116, mAs 270), which produced CTDIvol of 41.7 mGy and DLP of 935.8 mGy/cm. Crucially, the image quality with the optimized settings was maintained. This research validates a hybrid approach combining MC simulations and machine learning (ML) for effectively optimizing CT imaging parameters to reduce pediatric radiation exposure while preserving diagnostic image quality