(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: As automated planning systems become more broadly available, the most important questions of their impact center on resultant plan quality and efficiency gains. The purpose of this work was to assess the impact on plan quality variability and workforce efficiency as we implemented knowledge-based automated planning (KBP) across multiple disease sites and several hundred patients over three years.
Methods: The effects of KBP implementation were investigated in three sequential phases. Group A represented the manual planning reference just prior to implementation and were retrospectively re-planned to assess pre-autoplanning variability. For group B, manual and KBP plans were generated in parallel with treated plan selected by physician preference. In group C, KBP were generated as baseline for each case and manually-refined at planner discretion. Disease sites investigated were prostate(N(A)=53,N(B)=41,N(C)=218), prostatic fossa(N(A)=24,N(B)=32,N(C)=45), left lung SBRT(N(A)=27,N(B)=17,N(C)=43), right lung SBRT(N(A)=27,N(B)=19,N(C)=57), and head-and-neck(N(A)=52,N(B)=36,N(C)=141). The planner efficiency was quantified by plans/dosimetrist/day over all disease sites. To assess plan variability, we determined site-specific DVH parameters comparing OAR values for clinical and KBP plans: ?Dx=Dx,clinical-Dx,KBP mean and standard deviation. The per-patient comparison of both plans controls for individual patient anatomy on these parameters, while the standard deviation of ?Dx across the cohort measured plan variability. Significance testing between A and C utilized unpaired two-sided t-tests(p<0.001).
Results: We observed an 8.6%/year increase in planner productivity after the introduction of KBP (0.94?1.20 plans/dosimetrist/day). KBP reduced planning variability with significant differences in prostate plans for bladder ?V40Gy(2.4%±2.5%?-0.4%±2.2%,p<0.001), penile bulb ?Dmean(7.8%±7.6%?1.1%±3.8%,p<0.001) and rectum ?V40Gy(5.6%±5.2%?-0.7±2.9%,p<0.001), ?V65Gy(1.5%±1.8%?-0.2%±1.3%,p<0.001) and ?V75Gy(1.0%±1.0%?-0.1%±1.0%,p<0.001). Similar changes and standard deviation reductions were observed for prostatic fossa. Changes in lung were not significant. Head-and-neck reduced variability with significant differences in ?Dmean for right parotid(4.5Gy±5.7Gy?-0.2Gy±3.4Gy,p<0.001), cricopharyngeus(15.0Gy±8.0Gy?0.4Gy±3.5Gy,p<0.001), left parotid(3.5Gy±4.0Gy?-0.1Gy±2.8Gy,p<0.001), esophagus(4.2Gy±4.2Gy?0.2Gy±1.9Gy,p<0.001), larynx(5.8Gy±5.9Gy?0.3Gy±2.0Gy,p<0.001) and left cochlea(6.0Gy±7.8Gy?0.0Gy±2.2Gy,p<0.001).
Conclusion: KBP adoption reduced planning variability across multiple disease sites concurrent to substantially increased efficiency of planners.
Funding Support, Disclosures, and Conflict of Interest: KLM reports income for personal consulting and speaker honoraria from Varian Medical Systems.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
