Pet Image Based Intrabeam Range Verification And Delivery Optimization For Charged Particle Radiation Therapy

To develop a novel PET image based method to overcome long-standing technical challenges to significantly improve the accuracy of beam range verification and therapy targeting. Contrast to the conventional PET based beam range (BR) verification provided after the completion of each fraction, we proposed previously to acquire PET image with an initial “probing” beam to verify BR before the start of each fraction, which enables intra-fraction adaptive beam modification, if necessary, to improve the targeting. One major challenge to this method is the requirement of using low dose probing beam (with negligible therapy effect) that leads to low counts, low PET image quality, and increased error of BR measurement. To solve this problem, we investigated: 1) to acquire PET image with a portion of therapy beams selected from the original treatment plan to provide sufficiently high counts for imaging and accurate BR verification; 2) to minimize the risk of beam overshooting (i.e., beyond the furthest tumor boundary to be covered by therapy), the selected beams will target within the tumor (i.e., “mid-tumor” region), and the associated PET imaging and mid-tumor BR measurement may still provide necessary information to verify the treatment plan and, if needed, enable adaptive modification of the remaining beam deliveries through advanced optimization processes to ensure the planned dose distribution can be achieved accurately. Simulations were conducted: 1) a CT image generated digital human brain phantom with a 5 cm diameter tumor located near the center was irradiated with 104 MeV scanning proton beams (5 x 109 protons, ∼0.4 Gy) selected from the planned therapy beams; 2) PET image (2 mm resolution) with 43 cm in-plane and 6.4 cm axial field-of-view was acquired for ∼120 sec; 3) mid-tumor BRs were measured and compared to the predicted values; 4) an extra 2.5 cm diameter air (or bone) sphere was inserted inside the phantom along the beam path to alter the BR values; 5) above 1-3) steps were repeated; based on the shifts of measured and originally predicted beam ranges, the delivery of rest beams was modified. PET based BR measurement was accurate, with error of 0.4 mm (MEAN) and 1.6 mm (Standard Deviation). The difference between the delivered and planned doses (∼10 Gy) were evaluated for both simulated cases in step 4): without and with adaptive delivery, the Gamma-index pass rates were 75.8%, 82.6% and 93.9%, 98.7%, respectively. The Gamma-index criteria were 3% and 3 mm and including only target and Organ-at-risk voxels. The study demonstrated the feasibility and advantage of the new method to significantly improve the BR verification accuracy for adaptive beam delivery without compromising the dose distribution. This should lead to solving relevant fundamental technical problems and providing a practical solution to improve the targeting accuracy and therapy efficacy.