Vector-model-supported optimization in volumetric-modulated arc stereotactic radiotherapy planning for brain metastasis.
Journal:
Medical dosimetry : official journal of the American Association of Medical Dosimetrists
Published Date:
Mar 18, 2017
Abstract
Long planning time in volumetric-modulated arc stereotactic radiotherapy (VMA-SRT) cases can limit its clinical efficiency and use. A vector model could retrieve previously successful radiotherapy cases that share various common anatomic features with the current case. The prsent study aimed to develop a vector model that could reduce planning time by applying the optimization parameters from those retrieved reference cases. Thirty-six VMA-SRT cases of brain metastasis (gender, male [n = 23], female [n = 13]; age range, 32 to 81 years old) were collected and used as a reference database. Another 10 VMA-SRT cases were planned with both conventional optimization and vector-model-supported optimization, following the oncologists' clinical dose prescriptions. Planning time and plan quality measures were compared using the 2-sided paired Wilcoxon signed rank test with a significance level of 0.05, with positive false discovery rate (pFDR) of less than 0.05. With vector-model-supported optimization, there was a significant reduction in the median planning time, a 40% reduction from 3.7 to 2.2 hours (p = 0.002, pFDR = 0.032), and for the number of iterations, a 30% reduction from 8.5 to 6.0 (p = 0.006, pFDR = 0.047). The quality of plans from both approaches was comparable. From these preliminary results, vector-model-supported optimization can expedite the optimization of VMA-SRT for brain metastasis while maintaining plan quality.
Authors
Keywords
Adult
Aged
Aged, 80 and over
Brain Neoplasms
Computer Simulation
Cranial Irradiation
Decision Support Systems, Clinical
Female
Humans
Male
Middle Aged
Models, Biological
Pattern Recognition, Automated
Radiometry
Radiosurgery
Radiotherapy Dosage
Radiotherapy Planning, Computer-Assisted
Reproducibility of Results
Sensitivity and Specificity
Support Vector Machine