Near-miss Event in Lung Cancer Radiation Therapy Because of a Random Outlier of Target Volume.

Abstract

Accurate tumor and organ-at-risk contouring is a critical step in radiation therapy. Contouring procedures, both manual and automated, are prone to errors and to a large degree of interobserver and intraobserver variability. Radiation oncologists and/or medical physicists have to perform independent reviews of all contours for each patient before using them for treatment planning, which is a time-consuming, labor-intensive, and still not error-free process. We presented the tracing of a subtle near-miss event because of the presence of a random outlier in the contours of a lung tumor, very far from the actual gross tumor volume. The treatment plan was performed with an automated treatment engine using the volumetric-modulated arc therapy technique. Despite the implementation and adoption of systematic procedures of quality assurance in our clinical routine, the error crossed the barriers of peer review and was identified subsequently only in the step of pretreatment dosimetric verification. The error was corrected, and the patient was replanned before treatment initiation. In this case study, we showed that the random creation of false-positive target outliers may have a detrimental impact on patient dose when automated planning is performed. This risk is not negligible, and all strategies for improving the robustness of target segmentation should be pursued.