Session Item

This work explores the feasibility to use statistical process control (SPC) on various plan evaluation metrics to standardize the treatment planning decisions. Furthermore, these metrics were cross-correlated to assess which parameters are more suitable to quantify plan quality.

The baseline for the SPC was established with a pilot data set of 105 VMAT bronchial treatment plans. The centre line (CL), upper (UCL) , and lower control limit (LCL) were calculated for various parameters, such as dose coverage (DC), homogeneity index (HI), gradient index, PMU (plan normalized monitor unit used to predict the degree of plan modulation (Park 2019, doi: 10.1002/acm2.12589)), plan quality index (PQI, use to compare the achievement of planning goals) (Jornet 2014, doi.org/10.1016/j.radonc.2014.06.016), dose differences and Gamma-index of an independent dose calculation-based (cD.Pat-QA) and a measurement-based patient-specific plan verification (mPat.QA). The cross-calibration of 11 plan quality metrics was achieved with the multivariable correlation (Pearson) and the parameters with the strongest correlation were chosen to assess the plan quality.

Fig.1 shows the result of the SPC of four of the plan quality metrics that were examined. In general, a quick overview of the plan quality and the weakness of a plan compared to the others can be achieved. The analysis of the SPC showed some treatment plans have only a single weakness (plan 2, poor DC), while others have multiple problems (plan 61, high PMU, poor DC, and gamma-index). The result of the multivariable correlation analysis shows PMU, DC, and mPat-QA gamma index have multiple weak or strong correlations with most of the plan quality metrics that were evaluated. Plan quality index (PQI) didn´t correlate with any other plan quality metrics evaluated. This is similar to the findings of Jornet 2014, who found no correlation between plan complexity and PQI. 

The plan quality metrics PMU, DC, and mPat-QA gamma index are suitable to predict the treatment plan quality. Statistical process control can be used to indicate poor plan quality. Integration of this method in a TPS will allow comparing a newly calculated plan with the baseline shown in this study. By introducing this method, plans with lower quality may be identified earlier and more clearly during the workflow.