TABLE 5.

Summary of published works on KBP using RapidPlan module by Varian treatment planning system

Ref.	Training size	Validation size	Cancer site	Purpose
34	57	23	Lung (VMAT)	To develop an RP‐KBP model for malignant pleural mesothelioma for patients with two intact lungs
35	50	50	Head/Neck (IMRT)	To establish a threshold of improvements of treatment plans submitted to the clinical trials for head‐neck cancer (NRG‐HN001) through a multi‐institutional KBP model
36	104	25	NSCLC (VMAT)	To evaluate the feasibility of single institution KBP model as a dosimetric quality control for multi‐institutional clinical trials to RTOG 0617
37	30, 60	13	Liver (IMRT)	To study prediction capability of RP general model (Model G with 60 cases) versus RP‐specific model (Model S with 30 cases) and benchmark against clinical plans for liver IMRT
38	40	24	Esophageal cancer (VMAT)	To evaluate RP‐KBP for training models with plans optimized with a different treatment planning system (Eclipse and RayStation)
39	48	25	Prostate (VMAT)	To demonstrate the effectiveness of RP‐KBP for hypofractionated, multi‐target prostate patients
42	30	10	Head/Neck (VMAT, Proton)	To investigate whether RP based only on photon beam characteristics can be used to generate DVH‐predictions for proton therapy and whether this could correctly identify patients for proton therapy
55	35 (LR) 30 (HR)	10 HR and 10 LR VMAT	Prostate (VMAT)	To use KBP models created from helical tomotherapy plans [35 low‐risk (LR) and 30 high‐risk (HR)] for generating plans with different techniques (VMAT)
56	79	20	NPC (IMRT)	To investigate the improvements in planning efficiency and quality for patients with NPC IMRT treatments
57	82	45	GBM (VMAT)	To create an initial RP‐based KBP model for glioblastoma (GBM) and evaluate the planning efficiency of RP‐based planning against typical manual planning
59	70	24	Esophageal (VMAT)	To evaluate the performance of the RP module for esophageal cancer VMAT
60	45	25	Liver (VMAT)	To evaluate the performance of RP‐based optimized plan against manually created plans for hepatocellular cancer for clinical acceptability
61	38	10	Spine (SBRT)	To determine if RP is effective in improving the quality and efficiency of spine SBRT planning and evaluate the model for outliers
62	40 (P) 37 (C)	10 (P) 10 (C)	Prostate (IMRT) Cervical (VMAT))	To determine whether the RP module can efficiently produce IMRT and VMAT plans in the pelvic region in a single optimization and benchmark
63	43	60 (10, 7, 6, 7,13, 10, 7)	Prostate (VMAT)	To perform the multicentric validation of RP models on seven different centers and compared with corresponding manually optimized plans
65	30,30 60	15, 15	Head/Neck (VMAT)	To study whether differences in the composition of plan libraries influenced RP results for two patient groups using three different libraries and benchmark the model versus clinical plans. To evaluate the influence of model size
64	90	20	Head/Neck (VMAT)	To evaluate the potential of RP to automate the process for identifying the quality of patient‐specific plans through the correlation between predicted and achieved mean doses to the different OAR structures
66	20, 53, 60, 100, 123	>20	Prostate (VMAT)	To evaluate the performance of RP‐KBP at multiple radiation therapy departments and check its suitability for sharing the models.
67	80	70	Rectal (SIB)	To investigate the performance of RP‐KBP compared to manually optimized clinical plans for rectal SIB cases.
68	40	11 (Int.) 22 (Ext.)	Spine SBRT	To investigate whether a validated KBP model for NRG Oncology RTOG 0631 could be used as a retrospective clinical trial quality control tool
70	70	10	Head/Neck (VMAT)	To study the influence of outliers (Suboptimal plans) on the prediction of RP plans by adding suboptimal plans into a clean model with the increment of five plans.
74	83	20	Head/Neck (VMAT)	To assess the stability of RP generated plans for a different beam geometry, different management of bilateral structures, and dose fractionations. Two models were generated: a model separating ipsi‐and‐contralateral parotids and a model associating two parotids to a single structure.
75	51	30	Prostate (VMAT)	To investigate whether RP plans created through a single optimization (without any planner intervention during optimization) are clinically acceptable for prostate cancer patients
76	51	35	Cervical (IMRT)	To demonstrate an efficient method to train, refine (i.e., according to clinical trial dosimetric objectives), and validate the KBP model for an automated quality control system
71	60	20	Prostate (IMRT)	To investigate the role KBP can play in aiding a clinic's transition to a new treatment planning system
78	81	30	Pelvic (VMAT)	TO test if RP DVH estimation can be improved interactively through a closed‐loop evaluation process
117	81	10	Rectal (VMAT)	To study whether RapidPlan model trained on a technique (VMAT) and orientation can be used for another (30 IMRT plans)
72	27, 27	25, 25	Lung (VMAT) Prostate (VMAT)	To evaluate the performance of a model‐based optimization process for prostate and lung VMAT plans and evaluate its predictive power compared to manually created plans.
73	150	70	Breast (VMAT)	To evaluate the performance of a model‐based optimization process for whole breast VMAT

Abbreviations: DVH, dose–volume histogram; IMRT, intensity‐modulated radiation therapy; KBP, knowledge‐based planning; NPC, nasopharyngeal carcinoma; NSCLC, non‐small cell lung caner; RP, rapid‐plan; VMAT, volumetric‐modulated arc therapy.