Room: Exhibit Hall
Purpose: To collect and analyze the patient setup-time based on four different patient immobilization devices of 6MV photon-beam Volumetric-Modulated Arc-Therapy treatments on a linear-accelerator (VersaHD, Elekta, Stockholm, Sweden).
Methods: The patient setup-time of 5244 fractions (363 treatments) were collected (TABLE-1). The distributions were investigated along with Hypoexponential-distribution (EQUATION-1), a typical statistical model for analyzing the total time of completing a sequence of procedures.
Results: All histograms of observed setup-time (FIGURE-1) were asymmetric, surged rapidly to peaks (the most probable setup-time) which were lower than the means, and declined slowly afterwards. Standard-deviations were large, undesirably allowing negative setup-time in 95%-median samples being estimated according to Normal-distribution (TABLE-2). The data was not normally distributed as Jarque-Bera test reported <0.0001 p-value in all cases.Differences were <4 minutes across devices at the higher ends of the observed time ranges in 38%-, 68%- and 95%-median samples. In contrast, the observed variations of each device were more prominent, >4 and >10 minutes in the 68%- and 95%-median samples respectively. Hypoexponential-distribution offered better agreement with the observed histograms compared to Normal-distribution (solid-curves versus dotted-curves, FIGURE-1). The former also reported relatively less discrepancies to the ranges of observed setup-time variations in 38%-, 68%- and 95%-median samples than the latter (TABLE-2).
Conclusion: Regardless of immobilization devices used in these 5244 fractions, the setup-time means and standard-deviations were ineffective of suggesting the most probable setup-time and the range of setup-time variations. The setup-time was not normally distributed and better described by Hypoexponential-distribution.It was observed that setup-time variations were >10 minutes in 95% samples. It would contribute to the time variation of treatment room occupancy, causing challenges to schedule treatment timeslots efficiently. Such uncertainty is represented by the studied probability mode, which could be employed to devise a more efficient treatment scheduling scheme.
Modeling, Radiation Therapy, Pattern Recognition