Catching Errors With Trajectory Log Files

Abstract

To present our single-institution experience on catching errors with trajectory log file analysis. To present measurement based tolerances for treatment type specific machine parameters as well as site-specific imaging shifts that could be used as a beam shutdown signal in the event parameters exceed tolerances. The reported causes of failures, probability of occurrences (O), severity of effects (S), and the probability of the failures to be undetected (D) could be added to guidelines of FMEA analysis. From March 2013 through March 2014, 19 569 patient treatment fields/arcs were analyzed for the treatment of brain, chest wall, esophagus, head and neck, lung, pelvis, prostate, spine, extremities, and breast. This work included checking all 131 treatment delivery parameters for all treatment sites and all treatment delivery fractions. Radiation therapy system trajectory log files for all treatment field types as well as all imaging types were accessed, read in every 20ms, and every control point (total of 37 million parameters) compared to the physician approved plan in the planning system. Couch angle outlier occurrence: N=327, range=-1.7–1.2 deg; gantry angle outlier occurrence: N=59, range=0.09-5.61 deg, collimator angle outlier occurrence: N=13, range=-0.2–0.2 deg. Volumetric modulated arc therapy cases had slightly larger variations in mechanical parameters; MLC: 3D single control point fields have a maximum deviation of 0.04mm, 39 step and shoot IMRT cases have MLC -0.3 – 0.5mm deviations, all (1286) VMAT cases have -0.9–0.7mm deviations. Two possible serious errors were found: 1) A 4cm isocenter shift for the PA beam of an AP-PA pair, underdosing a portion of PTV by 25%; 2) Delivery with MLC leaves abutted behind the jaws as opposed to the midline as planned, leading to an underdosing of a small volume of the PTV by 25%, by just the boost plan. Due to their error origin, neither of these errors could have been detected by pretreatment verification. This work shows the importance of performing trajectory log file analysis for all treatment plans, and for all treatment fractions, preferably real time to catch typically undetected errors to avoid potentially adverse incidents.