Room: AAPM ePoster Library
Purpose: compare the plans generated by two automatic planning models generated under the same environmental conditions to see whether an interactive improved internal feedback system within the model can be established.
Methods: cases of pelvic tumor were selected, and IMRT(P0) generated by clinical model M0 was imported into Rapid Plan model base to establish DVH prediction model through automatic planning model in clinical use, and the new Rapid Plan model (M1) was generated through training and structure matching settings. Under unified conditions, 70 new IMRT plans Plan1(P1 group) were generated by M1, and Rapid Plan model (M2) was generated by P1 method. By the same method, 70 IMRT plans Plan2(P2 group) were generated by M2. Dosimetric differences between P1 and P2 were compared.
Results: generated by secondary model M2 is better than that of plan P1 generated by M1 model, and the difference is statistically significant (P<0.01). D2, D98 and HI in group P1 were better than those in group P2, and the difference was statistically significant (P<0.01). The MU of group P2 was significantly lower than that of group P1 in the two plans, and the difference was statistically significant (P<0.01). bladder V10 and left femoral head V40 in P2 group are superior to P1 group, with statistical difference (P<0.05). The bladder V30, V40, rectum V40, Dmean, left femoral head V10, V20, V40, Dmean, right femoral head V20, V30, Dmean in P1 group are all better than those in P2 group, the difference is statistically significant (P<0.05). Dosage parameters of other crisis organs, such as bladder V20, Dmean, rectum V10, V20, V30, right femoral head V10, V40, showed no significant difference (P>0.05).
Conclusion: interactive improved internal feedback system in the automatic planning model is feasible and meaningful.