Artificial Neural Network-based Model for Predicting Cardiologists' Over-apron Dose in CATHLABs

Area of Science:

• Medical Physics
• Radiological Protection
• Artificial Intelligence in Medicine

Background:

• Cardiologists in catheterization labs face significant occupational radiation exposure.
• Overlooking dosimeter use due to high-stress tasks necessitates predictive models for radiation doses.

Purpose of the Study:

• To develop and validate an artificial neural network (ANN) model for predicting over-apron radiation dose in cardiologists.
• To assess the efficacy of ANN compared to traditional models for radiation dose prediction.

Main Methods:

• An ANN model was trained using data from Monte Carlo simulations with varying X-ray spectra and tube orientations.
• Input features included dose area product (DAP), energy spectrum, and tube angulation.
• A multilayer perceptron neural network was utilized for prediction.

Main Results:

• The ANN model achieved high predictive accuracy with a correlation coefficient (R-value) of 0.95 and a root mean square error (RMSE) of 3.68 µSv.
• The ANN model significantly outperformed a linear regression model (R-value=0.48, RMSE=18.15 µSv).

Conclusions:

• ANN models offer a powerful and accurate tool for predicting occupational radiation doses in clinical settings.
• The developed model can enhance safety protocols and enable real-time exposure assessment for cardiologists.
• Future work should focus on integrating these models into real-time monitoring systems.