Statistical process control tools for setup reproducibility in quality control practices during nasopharyngeal carcinoma radiotherapy

This study aims to perform quality control (QC) practices for setup reproducibility during radiotherapy for nasopharyngeal carcinoma (NPC) using statistical process control (SPC) tools.

A total of 480 fractional images from 48 NPC patients with the first 10 fractions of the treatment were collected. In QC practices, setup errors were described using the histogram and normal curve, cumulative frequencies of absolute setup errors and 3D Euclidean Distance (Eu) were analyzed; the X ̅-S chart and process capability index (Cpk) with the variable Eu were utilized to identify whether the outlier occurred and to evaluate the QC process.

The translational setup error distributions were almost normal in Lateral, Longitudinal and Vertical directions and were narrower in Lateral and Vertical directions. Vertical translational errors and Eu with a larger magnitude sag appeared the most frequently. Between the couch sag and no sag, the Eu mean of 7 to 7 NPC patients with the same 3 patients was out of control and the standard deviation of Eu  of nil to 2 patients was outlier based on the X ̅-S chart, and the Cpk was 1.05 and 1.36 respectively, when the specification limit of translational errors was ±3 mm.

Daily imaging is necessary to increase setup reproducibility for NPC patients and more measures should be taken to facilitate quality assurance procedures. SPC is better applied to QC practices depending on the reliable data and the acceptable tolerance levels in further studies.