Room: Exhibit Hall
Purpose: PyPlanScoring (PPS) package is software that calculates Dose Volume Histogram (DVH) for plans generated by most commercially available Treatment Planning Systems (TPSs). PPS was built extending the dicompyler-core package [1] to perform structure volume oversampling using dose trilinear interpolation, mainly in small structures. PPS was built initially to evaluate plans submitted to the international radiotherapy plan competitions organized by Radiation Knowledge (RK). This work aims to validate PPS software using stress tests for its DVH algorithm and results from commercially available software.
Methods: The authors use published analytical DVH datasets [2]. Histograms of the accuracy of various DVH parameters (total volume, Dmax[Gy], Dmin[Gy], D99%[Gy], D95%[Gy], D5%[Gy], D1%[Gy], D0.03cc[Gy]. A clinical Lung SBRT with common targets, as well as, control volumes, rings. Absolute DVHs were generated using Eclipse and PlanIQ, and both were compared to PPS estimated DVHs.
Results: The first validation test uses high-resolution geometric shapes with varying dose grids. The number of points exceeding 3% difference from analytical is presented along with the range of deviations. Only, 8 points of D0 03cc[Gy] constraint exceeded 3%, range [-0.1% - 5.8%]. The second test used three subsampled axial contour spacings (1, 2, and 3mm) paired with the dose grid resolution, to emulate the clinical practice, 23 points exceeded 3% limit. DVH comparison with PlanIQ showed minimum, maximum and mean deviation of -7.1%, 2.4%, -0.1 %, and Eclipse's showed -3.3%, 1.5%, -0.1%, respectively.
Conclusion: The DVH calculation methodology implemented in PPS made it compatible with most Dose DICOM files exported from most TPS’s. The PPS was the plan evaluation software used in the 2017 RK international radiotherapy plan competition and found to be very accurately matching the outcomes obtained from most TPS's and analytical DVHs datasets. [1] https://github.com/dicompyler/dicompyler-core[2] http://dx.doi.org/10.1118/1.4923175
Not Applicable / None Entered.
Not Applicable / None Entered.