An interactive plan and model evolution method for knowledge‐based pelvic VMAT planning

Abstract

PurposeTo test if a RapidPlan DVH estimation model and its training plans can be improved interactively through a closed‐loop evolution process.Methods and materialsEighty‐one manual plans (P0) that were used to configure an initial rectal RapidPlan model (M0) were reoptimized using M0 (closed‐loop), yielding 81 P1 plans. The 75 improved P1 (P1+) and the remaining 6 P0 were used to configure model M1. The 81 training plans were reoptimized again using M1, producing 23 P2 plans that were superior to both their P0 and P1 forms (P2+). Hence, the knowledge base of model M2 composed of 6 P0, 52 P1+, and 23 P2+. Models were tested dosimetrically on 30 VMAT validation cases (Pv) that were not used for training, yielding Pv(M0), Pv(M1), and Pv(M2) respectively. The 30 Pv were also optimized by M2_new as trained by the library of M2 and 30 Pv(M0).ResultsBased on comparable target dose coverage, the first closed‐loop reoptimization significantly (P < 0.01) reduced the 81 training plans’ mean dose to femoral head, urinary bladder, and small bowel by 2.65 Gy/15.63%, 2.06 Gy/8.11%, and 1.47 Gy/6.31% respectively, which were further reduced significantly (P < 0.01) in the second closed‐loop reoptimization by 0.04 Gy/0.28%, 0.18 Gy/0.77%, 0.22 Gy/1.01% respectively. However, open‐loop VMAT validations displayed more complex and intertwined plan quality changes: mean dose to urinary bladder and small bowel decreased monotonically using M1 (by 0.34 Gy/1.47%, 0.25 Gy/1.13%) and M2 (by 0.36 Gy/1.56%, 0.30 Gy/1.36%) than using M0. However, mean dose to femoral head increased by 0.81 Gy/6.64% (M1) and 0.91 Gy/7.46% (M2) than using M0. The overfitting problem was relieved by applying model M2_new.ConclusionsThe RapidPlan model and its constituent plans can improve each other interactively through a closed‐loop evolution process. Incorporating new patients into the original training library can improve the RapidPlan model and the upcoming plans interactively.

Developing knowledge‐based planning for gynaecological and rectal cancers: a clinical validation of RapidPlan™

“…Before the use of KBP at our centre, GYN and rectum sites required on average 14 and 19 optimisations and 6.7 and 8.1 hours of optimisation time, respectively, to achieve a clinically acceptable plan. This is consistent with Hussein et al 11 and Wang et al 22 who describe high levels of planner intervention to ensure a truly optimal plan during conventional trial-and-error inverse planning, due to suboptimal or conflicting optimisation constraints. With the use of KBP for GYN and rectal sites, we are able to consistently achieve clinically acceptable plans in a single optimisation which has shown real-world (i.e.…”

“…In this present study, the number of patient cases in each RP model was considered relatively small with 22 and 32 in the GYN and rectum models, respectively, compared to other KBP models quoted in the literature such as Hussein et al 11 (37 cervix), Lian et al 13 (86 cervix), Wang et al 22 (81 Rectum), Wu et al 23 (80 rectum), Chanyavanich et al 21 (100 prostate) and Good et al 14 (132 prostate). In this study, small case numbers in each of the KBP models introduced variability of primary sites and prescription doses particularly in the GYN model.…”

“…However, successful plan generation from clinically acceptable plans for the validation cohort in a single optimisation is encouraging. This optimisation tool has positive implications for radiation oncology 22 detail that incorporating new and improved patient plans into the original KBP training library can improve model output for the previous and upcoming plans. With previous literature indicating the success of KBP-generated plans being dependent on the quality of the plans in the KBP model 11,24,25 , this remains an area for future analysis.…”

Knowledge‐based planning for intensity‐modulated radiation therapy: A review of data‐driven approaches

“…The pooled mean of reduction is 2.6 and 2.0 Gy for rectum and bladder, respectively. Incidentally, a more recent study of rectal cancer treatment published after the review articles were collected also resulted in an average reduction of 2.06 Gy in bladder mean dose …”

“…Incidentally, a more recent study of rectal cancer treatment published after the review articles were collected also resulted in an average reduction of 2.06 Gy in bladder mean dose. 79 To gain further understanding of the overall performance of KBP in prostate cancer planning, we have developed a visualization scheme to provide a summary view of nine KBP prostate studies that compared re-planned results with original clinically approved values. 20,25,27,34,35,56,57,60,63 As mentioned previously, the challenge of summarizing results across all studies lie in two aspects: (a) the results are based on different sample points of the DVH curve and measure changes along different directions (e.g., one study may use D35 while another use V65); (b) some of the studies report only the differences in DVH point metrics (e.g., D35 is reduced by 1.5) without providing the original clinically approved values.…”