Poster Poster Program Therapy Physics

Leveraging Artificial Intelligence for PET Shielding Thickness Estimation: A Comparative Analysis

Abstract
Purpose

Traditionally, shielding design in PET facility relies on manually interpolating semi-logarithmic transmission curves from NCRP 147 and AAPM TG 108. This process is time-consuming and varies depending on the user. This study introduces a new, fully automated Artificial Intelligence (AI)-based framework that estimates optimal lead and concrete shielding thicknesses in PET facilities using supervised machine learning regressors.

Methods

Three datasets were created by digitizing transmission curves from the NCRP and AAPM TG guidelines, and five model families, including Artificial Neural Networks (ANNs), Random Forest (RF), XGBoost, Extra Trees Regressor (ETR), Bagging Regressor (BR), and Gaussian Process Regressor (GPR), were trained on an 80/20 split. They were evaluated using metrics such as mean absolute error (MAE), root mean square error, coefficient of determination (R²), and statistical bias testing. All features were normalized using the z-score technique, with the mean and standard deviation calculated from the training set and then applied to the test set.

Results

The ETR outperformed the others in predicting properties of PET concrete, achieving an MAE of 0.157 cm and an R² of 0.9998. Regarding the PET lead, GPR performed excellent, with an MAE of 0.122 mm and an R² of 0.9999. ANN models demonstrated low bias in most cases, although unnormalized PET concrete ANNs exhibited small systematic offsets. Tree-based models exhibited notable bias despite having a high R², highlighting the importance of bias testing. The top model, ETR, was validated against expert calculations, showing close agreement with some discrepancies due to limitations of the manual method.

Conclusion

This AI-based approach reduces shielding design time from days to seconds, ensures compliance with guidelines, minimizes human error, and enables optimization of materials and structural loads, thereby advancing data-driven radiation protection for PET facilities.

People

Related

Similar sessions

Poster Poster Program
Jul 19 · 07:00
Python-Based Automation Framework for Annual Machine QA Data Archiving In Qatrack+

Annual water-tank measurements help ensure beam characteristics remain consistent with commissioning baselines. However, the lack of a standardized processing workflow and decentralized data storage makes it difficult to analyze...

Syed Bilal Ahmad, PhD
Therapy Physics 0 people interested
Poster Poster Program
Jul 19 · 07:00
User Expectations and Current Availability of HDR Brachytherapy Audits In Europe

The aim of this work was to evaluate the need to implement more dosimetric audits in high‐dose‐rate brachytherapy (HDR-BT) in Europe and to identify which characteristics such audits should meet according to users.

Javier Vijande, PhD Laura Oliver Cañamás
Therapy Physics 0 people interested