Large scale adoption of statistical process control (SPC) for volumetric modulated ARC therapy patient-specific quality assurance: A retrospective analysis on 1400 patients

Introduction

Statistical Process Control (SPC) is a tool widely used in industrial engineering for monitoring, controlling and, ideally, improving a process through statistical analysis. This technique can be applied to quality assurance (QA) practice in radiotherapy in order to characterize the treatment process variability.

Purpose

To apply SPC strategy to our routine VMAT pre-treatment verification QA.

Materials and methods

In the last years, more than 1400 patients were treated with Elekta VMAT at our institution. Plans were re-grouped according to treatment technique and disease sites: (1) 611 high-modulated complex SIB treatments (head-neck, brain, gynecological, ano-rectal; (2) 348 prostate treatments and (3) 445 liver, lung and other metastasis treated with extracranial stereotactic radiotherapy (SBRT). Groups 1 and 2–3 plans were optimized with Oncentra Masterplan (dual-arc) and Ergo++ (single arc) TPS. A total of 4030 dose measurements were performed with the PTW Seven29 array/Octavius phantom, both on coronal and sagittal planes. Doses comparison were evaluated using 3%/3 mm γ-analysis. Two metrics were evaluated: (a) points-percentage with g-value less than one (g%) and (b) mean gamma (gmean). Clinical specifications were: g% >90% and gmean <0.67. Shewhart charts were used to calculate the central (CL), upper control (UCL) and lower control limits (LCL). The processes capability was evaluated by means of Cpk indexes.

Results

γ pass-rate values significantly depend on plan complexity. For g%, CL and LCL were 93.9% and 89.4%, 99.1% and 96.6%, and 99.3% and 97.6%, for group 1, 2 and 3 respectively. For gmean, CL and UCL were 0.416 and 0.585, 0.361 and 0.557, and 0.304 and 0.427, for groups 1, 2 and 3 respectively. In all cases, the control limits are well within the clinical specifications. The Cpl/Cpu capability indices for g% and gmean resulted equal to 0.61 and 1.06 in group 1; 2.59 and 1.12 in group 2; 3.77 and 2.10 in group 3, respectively. g% process for group 1 was not capable at 3%/3 mm, but with 5% – 3 mm specification for g%, CL and LCL resulted 98.3% and 96.2% and Cpl was 2.84.

Conclusion

SPC is useful to quantifiably demonstrate the QA process conformance to clinical specifications.