2279 - Development of a Real-Time Dose Calculation Method for BNCT Based on Blood Boron Concentration

Purpose/Objective(s): In boron neutron capture therapy (BNCT), accurate dose adjustment based on a patient’s blood boron concentration immediately before irradiation is essential to minimize side effects and achieve precise dose modulation. The current dose-scaling method uses linear interpolation among three fixed boron concentrations (10, 25, and 40 ppm). This approach may result in irradiation exceeding the tolerance dose of normal tissues. However, dose recalculation takes 2–6 hours, making it impractical after determining the patient’s blood boron concentration. This study aimed to develop a real-time dose calculation method that adjusts the irradiation dose based on the patient’s blood boron concentration immediately before irradiation, thereby improving dose accuracy and reducing uncertainty in BNCT.

Materials/Methods: This retrospective study included 20 patients diagnosed with brain tumors. BNCT treatment plans were generated using the treatment planning system (TPS) with a 2 mm³ voxel size and 2% statistical uncertainty. For each patient, treatment plans were computed for blood boron concentrations ranging from 10 to 40 ppm in 1 ppm increments. Monitoring units (MU) were determined by calibrating the dose to 900 cGy in 1 cm³ of normal brain tissue.

Three interpolation methods were evaluated:

1. Three-point method: Plans at 10, 25, and 40 ppm.
2. Seven-point method: Plans at 10, 15, 20, 25, 30, 35, and 40 ppm.
3. Eleven-point method: Plans at 10, 13, 16, 19, 22, 25, 28, 31, 34, 37, and 40 ppm.

Each method employed either linear interpolation or fourth-order polynomial approximation, resulting in six patterns. MU values were compared with those obtained directly from the TPS, and differences were evaluated as percentages. The Games-Howell test was used to assess dose accuracy and variability. Computation time for each boron concentration was also measured.

Results: For linear interpolation, the mean MU error was 2.43±1.38% (three-point), 0.30±0.25% (seven-point), and 0.10±0.21% (eleven-point). For polynomial approximation, the mean MU error was 0.02±1.39% (three-point), 0.00±0.19% (seven-point), and 0.01±0.18% (eleven-point). Statistical analysis showed significant differences in accuracy among all methods, except that polynomial approximation demonstrated no significant differences among three, seven, and eleven points. Computation time varied from 5hours 30minutes (10 ppm) to 2hours 12minutes (40 ppm), with higher boron concentrations corresponding to shorter times. The mean total computation time for each method was 10hours 37minutes (three-point), 23hours 11minutes (seven-point), and 35hours 57minutes (eleven-point).

Conclusion: Polynomial approximation significantly improves dose accuracy while reducing both variability and overdose risk in BNCT. Considering both accuracy and computation time, the seven-point polynomial approximation is a practical and efficient choice for real-time dose adjustment.