Wyatt Jonathan J, Mohanraj Rekha, Mott Judith H
Northern Centre for Cancer Care, Newcastle upon Tyne Hospitals NHS Foundation Trust, Newcastle, United Kingdom.
Translational and Clinical Research Institute, Newcastle University, Newcastle, United Kingdom.
Adv Radiat Oncol. 2024 May 8;9(8):101538. doi: 10.1016/j.adro.2024.101538. eCollection 2024 Aug.
Multiple brain metastases can be treated efficiently with stereotactic radiosurgery (SRS) using a single-isocenter dynamic conformal arc (SIDCA) technique. Currently, plans are manually optimized, which may lead to unnecessary table angles and arcs being used. This study aimed to evaluate an automatic 4π optimization SIDCA algorithm for treatment efficiency and plan quality.
Automatic 4π-optimized SIDCA plans were created and compared with the manually optimized clinical plans for 54 patients who underwent single-fraction SRS for 2 to 10 metastases. The number of table angles and number of arcs were compared with a paired test using a Bonferroni-corrected significance level of < .05/4 = .0125. The reduction in treatment time was estimated from the difference in the number of table angles and arcs. Plan quality was assessed through the volume-averaged inverse Paddick Conformity Index (CI) and Gradient Index (GI) and the volume of normal brain surrounding each metastasis receiving 12 Gy (local V12 Gy). For a 5-patient subset, the automatic plans were manually adjusted further. CI and GI were assessed for noninferiority using a 1-sided test with the noninferiority limit equal to the 95% interobserver reproducibility limit from a separate planning study (corrected significance level < .05/[4 - 1] = .017).
The automatic plans significantly improved treatment efficiency with a mean reduction in the number of table angles and arcs of -0.5 ± 0.1 and -1.3 ± 0.2, respectively (±SE; both < .001). Estimated treatment time saving was -2.7 ± 0.5 minutes, 14% of the total treatment time. The volume-averaged CI and GI were noninferior to the clinical plans (both < .001), although there was a small systematic shift in CI of 0.07 ± 0.01. The resulting difference in local V12 Gy, 0.25 ± 0.04 cm, was not clinically significant. Minor manual adjustment of the automatic plans removed these slight differences while preserving the improved treatment efficiency.
Automatic 4π optimization can generate SIDCA SRS plans with improved treatment efficiency and noninferior plan quality.
使用单等中心动态适形弧（SIDCA）技术的立体定向放射外科（SRS）能够有效治疗多发脑转移瘤。目前，计划是手动优化的，这可能导致使用不必要的床角和弧。本研究旨在评估一种用于治疗效率和计划质量的自动4π优化SIDCA算法。
为54例接受单次分割SRS治疗2至10个转移瘤的患者创建了自动4π优化的SIDCA计划，并与手动优化的临床计划进行比较。使用Bonferroni校正的显著性水平<0.05/4 = 0.0125的配对检验比较床角数量和弧数量。根据床角数量和弧数量的差异估计治疗时间的减少。通过体积平均反帕迪克适形指数（CI）、梯度指数（GI）以及每个转移瘤周围接受12 Gy照射的正常脑体积（局部V12 Gy）评估计划质量。对于一个5例患者的子集，对自动计划进行进一步手动调整。使用单侧检验评估CI和GI的非劣效性，非劣效性界限等于来自另一项计划研究的95%观察者间再现性界限（校正显著性水平<0.05/[4 - 1] = 0.017）。
自动计划显著提高了治疗效率，床角数量和弧数量平均分别减少了-0.5±0.1和-1.3±0.2（±标准误；均<0.001）。估计节省的治疗时间为-2.7±0.5分钟，占总治疗时间的14%。体积平均CI和GI不劣于临床计划（均<0.001），尽管CI存在0.07±0.01的小系统偏移。局部V12 Gy的结果差异为0.25±0.04 cm，无临床显著性。对自动计划进行轻微手动调整消除了这些微小差异，同时保持了提高的治疗效率。
自动4π优化可以生成具有提高的治疗效率和不劣于手动计划质量的SIDCA SRS计划。
