作者: Dan Nguyen , David Thomas , Minsong Cao , Daniel O’Connor , James Lamb
DOI: 10.1118/1.4963212
关键词: Biomedical engineering 、 Monte Carlo method 、 Dosimetrist 、 Nuclear medicine 、 Medical imaging 、 Segmentation 、 Mathematics 、 Total variation denoising 、 Isocenter 、 Column generation 、 Dosimetry
摘要: Purpose Magnetic resonance imaging (MRI)-guided Co-60 provides daily and intrafractional MRI soft tissue for improved target critical organ tracking. To increase delivery efficiency, the system uses three sources at 120° apart, allowing up to 600 cGy combined dose rate isocenter. Despite potential tripling in output, creating a plan that all is considerably unintuitive. Here, authors computerize triplet orientation optimization using column generation, an approach was demonstrated effective integrated beam fluence noncoplanar therapies. achieve better quality without increasing treatment time, then solve map (FMO) problem while regularizing maps reduce number of deliverable MLC segments. Methods Three patients—one prostate, one lung, head neck boost (H&NBoost)—were evaluated this study. For each patient, beamlet doses were calculated Monte Carlo, under 0.35 T magnetic field, 180 equally spaced coplanar beams grouped into 60 triplets. The size 1.05 × 0.5 cm determined by leaf thickness step size. triplets selected generation algorithm. FMO formulated L2-norm fidelity term L1-norm anisotropic total variation regularization term, which allows controlling segments, hence with minimal degradation dose. authors’ Fluence Regularization Optimized Selection Triplets (FROST) plans compared against clinical (CLNs) produced experienced dosimetrist. PTV homogeneity, max dose, mean D95, D98, D99 evaluated. OAR doses, as well R50, defined ratio 50% isodose volume over planning investigated. Results The differ +0.04%, +0.07%, +0.25% prescription between methods. homogeneity virtually same values 0.8788 0.8812 (CLN). R50 decreased 0.67 comparing FROST CLN. On average, reduced Dmax Dmean OARs 7.30% 6.08% respectively. manual CLN processes required numerous trial error runs. on other hand human intervention. Conclusions Efficient MRI-guided therapy needs output multiple yet suffers from unintuitive laborious selection processes. Computerized improves both efficiency dosimetry. novel additional controls segments time.