GPU-based prompt gamma ray imaging from boron neutron capture therapy

Area of Science:

• Medical Physics
• Computational Imaging
• Nuclear Medicine

Background:

• Boron Neutron Capture Therapy (BNCT) is an effective cancer treatment.
• Accurate real-time imaging is crucial for effective BNCT.
• Prompt gamma ray imaging offers potential for treatment verification.

Purpose of the Study:

• To develop and validate a fast prompt gamma ray image reconstruction method for BNCT simulations.
• To leverage graphics processing unit (GPU) computation for accelerated image reconstruction.
• To assess the accuracy and speed of the GPU-based reconstruction algorithm.

Main Methods:

• Monte Carlo simulations were performed for BNCT.
• A modified ordered subset expectation maximization (OSEM) algorithm was implemented on a GPU.
• Image reconstruction was performed using fewer projections.
• Accuracy was evaluated using a phantom with defined Boron Uptake Regions (BURs) and Receiver Operating Characteristic (ROC) curve analysis.
• Computation times were compared between GPU and Central Processing Unit (CPU).

Main Results:

• GPU-based image reconstruction was 196 times faster than CPU-based reconstruction.
• ROC analysis showed varying accuracy for different BURs, with Area Under Curve (AUC) values ranging from 0.6726 to 0.8009.
• Feasibility of prompt gamma ray image reconstruction for BNCT was demonstrated.

Conclusions:

• GPU computation enables rapid reconstruction of prompt gamma ray images from BNCT simulations.
• This technique holds promise for enabling fast, in-treatment image reconstruction during BNCT.
• The study validates the use of GPU acceleration for prompt gamma ray imaging in BNCT.