Object and person tracking systems to enable extremity dosimetry in nuclear medicine using computational methods

Area of Science:

• Medical Physics
• Occupational Health
• Radiological Protection

Background:

• Nuclear medicine (NM) professionals are at risk of high radiation doses, especially to their hands.
• Traditional ring dosimeters for extremity dose monitoring have significant uncertainties.
• Accurate occupational dosimetry is crucial for worker safety in NM.

Purpose of the Study:

• To develop and assess computer vision and AI algorithms for precise tracking of radiation sources and worker extremities.
• To reduce uncertainties in occupational dose assessment in nuclear medicine.
• To provide a foundation for improved radiation safety monitoring.

Main Methods:

• Development of computer vision and artificial intelligence algorithms for object detection.
• Implementation of algorithms to track unshielded syringes and hands of NM professionals.
• Assessment of tracking accuracy and confidence using a defined hardware setup.

Main Results:

• High tracking accuracy achieved for hand detection (98%).
• Moderate tracking accuracy for syringe detection (57%).
• Demonstrated feasibility of using AI for real-time occupational monitoring.

Conclusions:

• AI-powered tracking offers a promising alternative to traditional dosimetry for NM workers.
• Improved positional data can significantly enhance the accuracy of Monte Carlo dose simulations.
• Further development could refine syringe tracking for comprehensive radiation safety solutions.