Quality Control Of High-Dose-Rate Brachytherapy: Treatment Delivery Analysis Using Statistical Process Control

Abstract

Purpose: Statistical process control (SPC) is a quality control method used to ensure that a process is on-target and operating with minimal variation. The focus of this study is to determine if SPC is a viable technique for evaluating the proper operation of a high-dose-rate (HDR) brachytherapy treatment delivery system. Materials and Methods: Thomson-Nielsen MOSFET Dosimetry System (MOSFET model 502RD) was used as the dosimeter. MOSFET dosimeters (MD) were calibrated against a Farmer-type ionization chamber at a distance of 1.6 cm from 192Ir in water. A surrogate prostate patient was developed utilizing Vyse Ordance™ gelatin. Needle placement was performed according to our prostate HDR clinical protocol. A total of 10 MD were placed from prostate base to apex. Three detector trains of three MD each, spaced 2.5 cm apart, were inserted into two detector holders at the prostate periphery as well as in the urethra. The remaining MD was placed at the apex on the periphery. CT guidance was used to accurately position the first detector in each train at the base. The plan consisted of 12 needles with 129 dwell positions spaced 0.5 cm apart, delivering a prescribed peripheral dose of 200 cGy. Marks were placed on the needles at the template, and the distance from the template to the needle hub was recorded to allow accurate repositioning of each needle. Sixteen accurate treatment trials were delivered as planned. Subsequently a number of treatments were delivered with errors introduced. The errors were: wrong patient (i.e., a clinical plan for a different patient with 12 needles), wrong source calibration (3 & 7 day source decay inaccuracy), wrong sequence (i.e. two needles connected to incorrect location in turret), single needle displaced inferiorly 5mm, and entire implant displaced 2mm and 4mm inferiorly. Each MD constituted an evaluation subgroup size of one (bias correction factors d2 & d3 determined for n = 2 for consistency with established SPC methodology). The range was therefore determined by the difference between sequential treatment delivery measurements. Two process behavior charts (PBC), an average and range chart, were developed for each MD. Results: There were 4 false positives resulting from 160 measurements from 16 accurately delivered treatments. Generally speaking, data points resulting from accurately delivered treatments were well within PBC limits. The PBC indicate that measurements made at the base and apex were much more sensitive to treatment delivery errors. Each of the errors introduced was correctly identified by either the average or range PBC for measurements made at the base (2 out of 3 locations) and apex (all 4 locations). The PBC for measurements made at mid-gland (3 locations) were less sensitive to errors in treatment delivery. The only errors detected at these locations were “wrong patient” and “wrong source calibration.” Conclusions: SPC is a viable methodology to assess the quality of HDR treatment delivery. Further development is necessary to determine the most effective positioning of the MD for detecting potential treatment delivery errors.