Catching errors with patient-specific pretreatment machine log file analysis

Dharanipathy Rangaraj; Mingyao Zhu; Deshan Yang; Geethpriya Palaniswaamy; Sridhar Yaddanapudi; Omar H Wooten; Scott Brame; Sasa Mutic

doi:10.1016/j.prro.2012.05.002

Catching errors with patient-specific pretreatment machine log file analysis

Pract Radiat Oncol. 2013 Apr-Jun;3(2):80-90. doi: 10.1016/j.prro.2012.05.002. Epub 2012 Jun 7.

Authors

Dharanipathy Rangaraj¹, Mingyao Zhu², Deshan Yang², Geethpriya Palaniswaamy², Sridhar Yaddanapudi², Omar H Wooten², Scott Brame², Sasa Mutic²

Affiliations

¹ Department of Radiation Oncology, Washington University School of Medicine, St Louis, Missouri. Electronic address: drangaraj@swmail.sw.org.
² Department of Radiation Oncology, Washington University School of Medicine, St Louis, Missouri.

PMID: 24674309
DOI: 10.1016/j.prro.2012.05.002

Abstract

Purpose: A robust, efficient, and reliable quality assurance (QA) process is highly desired for modern external beam radiation therapy treatments. Here, we report the results of a semiautomatic, pretreatment, patient-specific QA process based on dynamic machine log file analysis clinically implemented for intensity modulated radiation therapy (IMRT) treatments delivered by high energy linear accelerators (Varian 2100/2300 EX, Trilogy, iX-D, Varian Medical Systems Inc, Palo Alto, CA). The multileaf collimator machine (MLC) log files are called Dynalog by Varian.

Methods and materials: Using an in-house developed computer program called "Dynalog QA," we automatically compare the beam delivery parameters in the log files that are generated during pretreatment point dose verification measurements, with the treatment plan to determine any discrepancies in IMRT deliveries. Fluence maps are constructed and compared between the delivered and planned beams.

Results: Since clinical introduction in June 2009, 912 machine log file analyses QA were performed by the end of 2010. Among these, 14 errors causing dosimetric deviation were detected and required further investigation and intervention. These errors were the result of human operating mistakes, flawed treatment planning, and data modification during plan file transfer. Minor errors were also reported in 174 other log file analyses, some of which stemmed from false positives and unreliable results; the origins of these are discussed herein.

Conclusions: It has been demonstrated that the machine log file analysis is a robust, efficient, and reliable QA process capable of detecting errors originating from human mistakes, flawed planning, and data transfer problems. The possibility of detecting these errors is low using point and planar dosimetric measurements.