Home About us Editorial board Ahead of print Current issue Search Archives Submit article Instructions Subscribe Contacts Login 

 Table of Contents  
Year : 2022  |  Volume : 18  |  Issue : 2  |  Page : 432-437

Study on the feasibility of quality improvement for automatic plans based on rapid plan model

1 Department of Radiotherapy, Shandong Cancer Hospital and Institute, Shandong First Medical University and Shandong Academy of Medical Sciences, Shandong Province, China
2 Department of Radiotherapy, Qilu Hospital of Shandong University, Jinan, China
3 Department of Hematology, Qilu Hospital of Shandong University, Jinan, China

Date of Submission10-Jan-2022
Date of Decision25-Jan-2022
Date of Acceptance28-Jan-2022
Date of Web Publication06-May-2022

Correspondence Address:
Changsheng Ma
Department of Radiotherapy, Shandong Cancer Hospital and Institute, Shandong First Medical University and Shandong Academy of Medical Sciences, Jinan 250117, Shandong Province
Shuang Yu
Department of Hematology, Qilu Hospital of Shandong University, Jinan 250012
Login to access the Email id

Source of Support: None, Conflict of Interest: None

DOI: 10.4103/jcrt.jcrt_65_22

Rights and Permissions
 > Abstract 

Objective: To study whether an interactive improved internal feedback system with the model can be established, we compared the plans generated by two automatic planning models.
Methods: Seventy cases of pelvic patients were selected. Intensity-modulated radiation therapy (IMRT) plans (P0) generated by the clinical model (M0) were imported into the Rapid plan model to establish a dose-volume histogram. The predicted model through automatic planning model in clinical, and the new rapid plan model (M1) was generated by training and structure matching settings. The 70 new IMRT plans (P1) were generated by M1, and the new rapid plan model (M2) was trained by P1. In this same method, 70 IMRT plans (P2) were generated by M2. Dosimetric differences between P1 and P2 were then compared and analyzed.
Results: For the model parameters, R2 and X2 in P2 were higher than those in P1, and the CD values of the bladder, right femoral head, and rectum in P1 were higher than those of corresponding organs in P2. The studentized residual (SR) value of the bladder and SR and difference of estimate values of the left femoral head and right femoral head in P1 were lower than P2. In planning, (D2, D98, and HI) P1 were better than P2 (P < 0.01); the bladder V10 and left femoral head V40 in P2 were lower than in P1 by 0.08% and 0.15%, respectively (P < 0.05); others in P2 were higher than those in P1 (P < 0.05) except the bladder V20, Dmean, rectum V10, V20, V30, right femoral head V10, and V40; and the MUs of P2 was lower than that of P1 for 132.2 (P < 0.05).
Conclusion: The stability of M2 is stronger than that of M1. Therefore, the interactive improved internal feedback system within the model of “plan-model-plan-model” is feasible and meaningful.

Keywords: Cervix cancer, dose, intensity-modulated radiation therapy, knowledge-based model, rapid plan

How to cite this article:
Li K, Ma C, Zhang X, Tao C, Ma C, Yu S. Study on the feasibility of quality improvement for automatic plans based on rapid plan model. J Can Res Ther 2022;18:432-7

How to cite this URL:
Li K, Ma C, Zhang X, Tao C, Ma C, Yu S. Study on the feasibility of quality improvement for automatic plans based on rapid plan model. J Can Res Ther [serial online] 2022 [cited 2022 Oct 1];18:432-7. Available from: https://www.cancerjournal.net/text.asp?2022/18/2/432/344877

 > Introduction Top

Optimizing the radiation treatment plan is a process of trial and error that obtains a satisfactory dose distribution between the planning target volume (PTV) and organs at risks (OARs). Under the condition that the PTV and dosage are determined, the quality of plans has a great relationship with the angle and optimization conditions given by the physicist. Due to physicists and their different experiences and levels in different hospitals, there are significant differences in the radiotherapy plans designed for the same case. The radiotherapy planning design of “knowledge-based radiation therapy” (KBRT), which was proposed by the research team of Duke University,[1],[2],[3] is a good method to solve this problem.

KBRT, as a typical representative of artificial intelligence and big data application in radiotherapy, has been well-known, accepted, and recommended by the industry.[4],[5] Presently, each mainstream planning system also has its own KBRT. For example, the Eclipse planning system of Varian Company has launched its automatic planning system rapid plan, which is commonly referred to as rapid planning or automatic planning. Eclipse 13.5 v. 13.5 and later can turn on this feature. This study uses Eclipse v. 13.6.

The newly established model needs further examination and analysis to eliminate the problems caused by data import, PTV and organ structure matching, prescription dosage, and other steps.[6] The following parameters are mainly checked: (1) residual scatter plot reflects the difference between the actual value and predicted value of dose-volume histogram (DVH); (2) Regression curve reflects the relationship between main geometric features and DVH; (3) The box map of the geometric distribution of organs at risks reflects the anatomical features used in the model training plan; (4) The distribution of DVH in the field reflects the correlation between the real value and the predicted value of DVH in the field; (5) training log files that record statistical characteristics of fitting results. To detect outliers or strong influence points, set the following thresholds: Cook's distance value (CD) >4; modified Z-score (MZ) >3.5; studentized residual (SR) >3; areal difference of estimate (DA) >3.[7],[8],[9] Therefore, We established a new rapid planning model and explored its clinical practicability based on the above.

 > Materials and Methods Top

Basic information

Intensity-modulated radiation therapy (IMRT) planning for 70 pelvic tumors in Shandong Tumor Hospital were randomly selected, regardless of age, sex, disease type, and stage. All plans were performed using 6MV FF X-rays at a 400 MU/minute dose rate. Different types of accelerators were allowed to be selected for different cases, while the requirements for the type and radiation fields of the same case were the same. Since this is a retrospective and offline study, it is unnecessary to need ethics approval; neither does it contain written informed consent from participants [Table 1].
Table 1: Statistical analysis of two groups of models (M1, M2)

Click here to view

Model establishment and generation plan

Using the existing rapid plan clinical model (M0) generated by the clinical plan and the DVH prediction model, the structural conditions are uniformly matched, the plan optimization is automatically completed, and the completed IMRT plan is named Plan0 (P0). First, P0 of 70 cases was re-imported into the Varian rapid plan model library. Then, training and structure matching settings were conducted to generate a new rapid plan model (M1) as the primary model. Structural matching is limited to PTV, bladder, rectum, and femoral heads.

The newly generated automatic planning model, M1, was selected to continue preparing the new IMRT plan for 70 cases, 6 MV FF X-ray, and 400 MU/minute dose rates were also selected. We will name the new IMRT plan generated by the automatic plan model, M1 Plan1 (P1 group). Through the same method, Plan1 was used to establish a new rapid plan model, M2, as a secondary model, and the same training and structure matching as M1 was conducted. Using model M2, the same conditions as Plan1 were selected to generate a new IMRT plan, Plan2 (P2) [Figure 1].
Figure 1: Flowchart of planning method

Click here to view

We will set rapid plan models M1 and M2 in a unified way, to reduce human interference in automatic plan optimization and ensure the consistency of optimization conditions. In the optimization process, no matter to the extent each target condition is reached, or whether a single plan optimization result is well, no human adjustment will be made in the process of plan completion. The same normalization method will be adopted after the plan is completed.


P1 generated by M1 was taken as the first group, and P2 generated by M2 as the second group. The automatic planning of the same case keeps the same algorithm, angle, and prescription dose. However, the plan between different cases may have different angles and doses (a single dose of 1.8 Gy, or 2.0 Gy, 50 Gy, or 50.4 Gy). P0 generated by the clinical model M0 was not used because the clinical plan established M0. Although we set the same conditions and did not make manual adjustments because the manual plan established the model. The P0 generated by M0 was inevitably influenced by human fine-tuning factors and tended to people's ideal dose index.

Plan evaluation

DVH statistics were used to evaluate the planned organ-threatening parameters of each group, bladder/rectum/femoral heads V10, V20, V30, V40, and Dmean. The MU (monitor unit, sum of all field hops in a certain plan), maximum dose D2, minimum dose D98, target dose HI (Homogeneity Index), and CI (Conformability Index) of the two plans were evaluated. The maximum exposure dose, D2, is defined as the exposure dose received by 2% of PTV. The lowest exposure dose, D98, is defined as the exposure dose received by 98% of PTV. HI = (D2-D98)/Dpres, where Dpres is the prescription dose. The smaller the HI value, the better the uniformity of PTV.[3] CI = (VT, ref/vt) × (vt, ref/Vref), vt, ref is the PTV enclosed by the reference isodose line, vt is the PTV, Vref is the total volume under the reference isodose line.[4]

Statistical analysis

The Statistical Package for the Social Sciences (SPSS) 19.0 (SPSS Inc., Chicago, IL, USA) was used for data analysis. DVH data were extracted from Eclipse and imported into the analysis software in a tabular format. The comparison mean-paired sample t-test was used, and the difference was statistically significant (P < 0.05).

 > Results Top


R2and X2

We also need to analyze from the inside of the model to see which model is better and compare the targets between P1 and P2. Like the regression coefficient R2 of the bladder in two models, R2 of the bladder in M2 is 0.935, higher than that in M1 (0.892). As shown in the [Figure 2], the R22 value of M2 was higher than M1, indicating that the better convergence of M2, the more stable the result, and the better robustness. Furthermore, X2 in M2, the value of X2 in both the single organ and whole body were higher than that of X2 in M1. Therefore, it can be understood that M2 is more likely interrelated than M1, closer to our requirements for model setting, reflect our requirements for model training, or be closer to the requirements of the clinical plan for model generation, and the better its stability [Figure 2].
Figure 2: R2 and X2 for M1 and M2

Click here to view

Geometric outliers

There is the concept of a strong influence point in the model, which CD generally identifies as the point that strongly influences the regression model. The strong influence point is not necessarily the abnormal point, but it affects the result of the regression model. CD indicates the strong influence point in the regression model. The larger the value, the greater its influence on the model. It has a threshold setting. Compared with the two models, the CD (CD1) values of the bladder, right femoral head, and rectum in M1 were higher than those of corresponding organs in M2 because we reported earlier that the human positive factor in P0 was higher than P1. Combined with the fact that CD1 is greater than CD2, the effect of CD1 was greater than that of CD2. This is why some indexes in P1 were higher than those in P2. The geometric outliers MZ in the model MZ (MZ1) of the bladder, left femoral head, and rectum in M1 was higher than those in M2. MZ represents the geometric characteristics and other geometric outliers of the same structure in the model [Table 2].
Table 2: Statistical analysis of two groups of plans (P1, P2) generated by two models (M1, M2)

Click here to view


We can also use SR and DA to check whether a plan is over-optimized. SR = standard deviation of residuals. If one plan is over-optimized, the dose distribution will be much better than another plan. The SR values of bladder and femoral heads in M1 were lower than the corresponding values in M2; the DA values of bladder and rectum in M1 were higher than the corresponding values in M2; the DA values of femoral heads in M2 were higher than the corresponding values in M1, and the DA predicted the actual DVH [Figure 3].
Figure 3: Value of CD, MZ, SR, and DA for M1 and M2

Click here to view

Comparison of target dose parameters planned by two models

[Table 1] shows that the CI of P2 generated by M2 was better than that of P1 generated by M1, and the difference was statistically significant (P < 0.01). Likewise, D2, D98, and HI in P1 were better than those in P2, and the differences were all statistically significant (P < 0.01) [Figure 4].
Figure 4: Dose-volume histogram of organs at risks s for P1 and P2

Click here to view

Comparison of dose parameters of organs at risks planned by two models

Compared with various parameters of OARs planned by the two models, all data are shown in [Table 3] and [Figure 5]. Dosage parameters of some OARs, such as bladder V20, Dmean, rectum V10, V20, V30, right femoral head V10, V40, showed no significant differences (P > 0.05).
Table 3: Change Value of two groups of plans (P1, P2)

Click here to view
Figure 5: Mean dose of organs at risks s for P1 and P2

Click here to view


As shown in [Table 3], the MU of P2 was significantly lower than that of P1 in the two plans, and the difference was statistically significant (P < 0.01).

 > Discussion Top

KBRT has proved its advantages and reliability and has been accepted for planning in practical work. It mainly includes the establishment of the DVH prediction model and model training.[10],[11],[12],[13] The establishment of the DVH prediction model is to calculate the Geometry-Based Expected Dose (GED) of each organ in the provided treatment plan. GED evaluates the volume of the PTV and OARs, the distance between them, and the dose distribution at this distance. Training the model uses the principal component analysis method to conduct regression analysis on the planned GED and DVH to obtain the DVH and geometric condition-related parameters of each anatomical structure. In designing a new plan, the DVH prediction model calculates the possible DVH fluctuation range of the plan result through the correlation parameters according to the mutual positional relationship between the PTV on the patient image and the normal tissue and selects its lowest dose limit as the target optimization condition.

This study discusses whether a “plan-model-plan-model” internal feedback system with interactive improvement can be formed in the closed-loop state.[14],[15],[16] The results showed that the P2 was better than P1 in MU, CI, bladder V10, and left femoral head V40. In other aspects, the difference was also minimal. The left femoral head V10, V20, V30, and right femoral head V20, V30, Dmean index change was more obvious; bladder V30, V40, rectum V40, Dmean, and other indicators change was little; especially bladder V40, rectum V40, Dmean index change was <0.5%. Since plan optimization is a multi-objective optimization, there will be some improvement of indicators. Some indicators have not improved or even become worse but generally can be controlled within the required range and if changes are floating within the required range.

The larger the Chi-square value, the smaller the probability of independence and the greater the probability of correlation. In this study, the Chi-square value of the organ index of M2 is greater than that of M1. It showed that the closer it is to the set requirements for the model, the better it can reflect the plan we use to train the model. The M2 is relatively stable, and it can better reflect the requirements of the clinical plan.

The reason is that KBRT integrates the past treatment experience into the treatment of new patients. It uses many previous similar plans to train fitting models.[17] The verified model will be used to evaluate the anatomical structure and prescription dosage of new patients, especially the distance and interlacing between PTV and OARs. According to this, the model predicts the target parameters of DVH that the case may reach. The plan used to build the model affects the use effect of the model. In addition, due to the individual differences between cases and different clinical requirements, physicists sometimes have to make manual fine adjustments to achieve the effect of excellence. In establishing the experimental model, M1 was established based on P0, then P1 was generated, and then, M2 was established through P1 to generate P2. From M1, P0 to M2, P2, the artificial influence factor of the automatic planning model is gradually weakened. Furthermore, a positive factor in planning optimization is weakened, or the influence of this positive factor is an increasing trend for the forward model.

Varian KBRT divided the OARs into four parts: (1) shooting into the field and scattering; (2) The exposure dose between leaves is lower; (3) In the shooting field, the irradiated dose has obvious influence; (4) PTV overlap, and the irradiated dose is equivalent to the PTV dose. This part has the most important influence on the PTV dose distribution. Establishing the Rapidplan model involves importing the image, outline, dose, and DVH of case intensity adjustment plan into the Eclipse-rapid plan planning system for regression analysis of the DVH curves of various OARs to create a DVH prediction model. After matching, a new plan is made using the established model; the DVH prediction model will automatically generate the irradiation dose-volume range for tissues and organs and give the optimal DVH curve satisfying the current plan, becoming the target center of the dose limit value for the subsequent optimization. This calculation method is a two-dimensional algorithm, while the plan involves three-dimensional images, so the two-dimensional algorithm has its limitations in calculating the three-dimensional volume and dose distribution.[18],[19] Therefore, it may be more appropriate to use a three-dimensional calculation method. The improvement of the algorithm should play a more critical and core role in improving the rapid plan model.

The sourcing plan on which M2 is built is in a nonadvantageous state compared with the source plan used by M1, which we want to avoid but exist. However, even under such circumstances, the plan generated by M2 can be improved in some aspects. Simultaneously, the stability of M2 is better than M1.

 > Conclusion Top

It can be considered that the interactive improved internal feedback system within the model of the “plan-model” is feasible and meaningful. Simultaneously, for the model used clinically, we should pay attention to the continuous improvement of the model by using the excellent clinical plan completed in the later stage of the model.


This work was supported by Natural Science Foundation of Shandong Province (ZR2019MH136, ZR2020QA089), Project funded by China Postdoctoral Science Foundation (2019M652356), The National Nature Science Foundation of China (81800156, 81974467, 12005119).

Financial support and sponsorship


Conflicts of interest

There are no conflicts of interest.

 > References Top

Yang Y, Ford EC, Wu B, Pinkawa M, van Triest B, Campbell P, et al. An overlap-volume-histogram based method for rectal dose prediction and automated treatment planning in the external beam prostate radiotherapy following hydrogel injection. Med Phys 2013;40:011709.  Back to cited text no. 1
Varian Medical Systems. Eclipse Photon and Electron Instructions for Use [M]. Palo Alto, CA; 2014. p. 183-213.  Back to cited text no. 2
Varian Medical Systems. Eclipse Photon and Electron Reference Guide [M]. Palo Alto, CA; 2014. p. 263-348.  Back to cited text no. 3
Gown AM. Current issues in ER and HER2 testing by IHC in breast cancer. Mod Pathol 2008;21 Suppl 2:S8-15.  Back to cited text no. 4
Bragg CM, Conway J, Robinson MH. The role of intensity-modulated radiotherapy in the treatment of parotid tumors. Int J Radiat Oncol Biol Phys 2002;52:729-38.  Back to cited text no. 5
Chanyavanich V, Das SK, Lee WR, Lo JY. Knowledge-based IMRT treatment planning for prostate cancer. Med Phys 2011;38:2515-22.  Back to cited text no. 6
Zhu X, Ge Y, Li T, Thongphiew D, Yin FF, Wu QJ. A planning quality evaluation tool for prostate adaptive IMRT based on machine learning. Med Phys 2011;38:719-26.  Back to cited text no. 7
Lian J, Yuan L, Ge Y, Chera BS, Yoo DP, Chang S, et al. Modeling the dosimetry of organ-at-risk in head and neck IMRT planning: An intertechnique and interinstitutional study. Med Phys 2013;40:121704.  Back to cited text no. 8
Moore KL, Brame RS, Low DA, Mutic S. Experience-based quality control of clinical intensity-modulated radiotherapy planning. Int J Radiat Oncol Biol Phys 2011;81:545-51.  Back to cited text no. 9
Appenzoller LM, Michalski JM, Thorstad WL, Mutic S, Moore KL. Predicting dose-volume histograms for organs-at-risk in IMRT planning. Med Phys 2012;39:7446-61.  Back to cited text no. 10
Good D, Lo J, Lee WR, Wu QJ, Yin FF, Das SK. A knowledge-based approach to improving and homogenizing intensity modulated radiation therapy planning quality among treatment centers: An example application to prostate cancer planning. Int J Radiat Oncol Biol Phys 2013;87:176-81.  Back to cited text no. 11
Yuan L, Ge Y, Lee WR, Yin FF, Kirkpatrick JP, Wu QJ. Quantitative analysis of the factors which affect the interpatient organ-at-risk dose sparing variation in IMRT plans. Med Phys 2012;39:6868-78.  Back to cited text no. 12
Fogliata A, Wang PM, Belosi F, Clivio A, Nicolini G, Vanetti E, et al. Assessment of a model based optimization engine for volumetric modulated arc therapy for patients with advanced hepatocellular cancer. Radiat Oncol 2014;9:236.  Back to cited text no. 13
Djajaputra D, Wu Q, Wu Y, Mohan R. Algorithm and performance of a clinical IMRT beam-angle optimization system. Phys Med Biol 2003;48:3191-212.  Back to cited text no. 14
Narayanan VK, Vaitheeswaran R, Bhangle JR, Basu S, Maiya V, Zade B. An experimental investigation on the effect of beam angle optimization on the reduction of beam numbers in IMRT of head and neck tumors. J Appl Clin Med Phys 2012;13:3912.  Back to cited text no. 15
Kim SH, Kang MK, Yea JW, Kim SK, Choi JH, Oh SA. The impact of beam angle configuration of intensity-modulated radiotherapy in the hepatocellular carcinoma. Radiat Oncol J 2012;30:146-51.  Back to cited text no. 16
Hao W, Fan J, Haizhen Y, Yibao Z. Treatment of statistical outliers and their dosimetric effects in varian rapid plan model training. Chin J Med Phys 2016;3:649-53.  Back to cited text no. 17
Nwankwo O, Mekdash H, Sihono DS, Wenz F, Glatting G. Knowledge-based radiation therapy (KBRT) treatment planning versus planning by experts: Validation of a KBRT algorithm for prostate cancer treatment planning. Radiat Oncol 2015;10:111.  Back to cited text no. 18
Altinok P, Tekçe E, Karaköse F, Berk K, Kalafat Ü, Kiziltan HS, et al. Dosimetric comparison of modern radiotherapy techniques for gastric cancer after total gastrectomy. J Cancer Res Ther 2020;16:S133-7.  Back to cited text no. 19


  [Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]

  [Table 1], [Table 2], [Table 3]


Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
Access Statistics
Email Alert *
Add to My List *
* Registration required (free)

  >Abstract>Introduction>Materials and Me...>Results>Discussion>Conclusion>Article Figures>Article Tables
  In this article

 Article Access Statistics
    PDF Downloaded37    
    Comments [Add]    

Recommend this journal