- Email: [email protected]
OC-0254: Prospective validation of independent DVH prediction for QA of automatic treatment planning
S128 ESTRO 36 _______________________________________________________________________________________________ errors below 2.5Gy. For proton H&N plans, a dataset size of at least 173 plans resulted in all mean errors below 2.5Gy. Dataset sizes are shown in Table 1. Shown visually in Figure 1, using predictive modeling of the plan outcome, re-planning a lung SBRT case resulted in improved dose to critical structures while maintaining coverage to the PTV, compared to the clinically-developed and treated plans. Error Target (Gy) 5 4 3 2.5 2
deteriorate them by more than 1Gy/1% compared to the original plan. Given the 5 parameters of interest and the 90% criteria, (1- (0.9)5 )≈40% patients were expected to fall outside the prediction range. Results Figure 1 shows the results of the model validation.
Lung dataset size 16 46 61 69 74 HN Proton size
169 130 153 173 192
HN Photon size 68 136 145 121 136 Table 1: Plan datasets required for desired dose accuracy.
Figure 1 Comparison of clinical plan developed without (A) and with (B) predictive modeling. Conclusion We have demonstrated the ability to predict dosimetric indices. These results have clinical implications that extend from decision making to planning workflow improvement to quality improvement. OC-0254 Prospective validation of independent DVH prediction for QA of automatic treatment planning Y. Wang1, B.J.M. Heijmen1, S.F. Petit1 1 Erasmus MC - Cancer Institute, Radiation Oncology, Rotterdam, The Netherlands Purpose or Objective In our institute, fully automated, knowledge-based treatment planning is used in routine clinical practice. For the majority of patients, this is expected to result in high quality treatment plans. However, technical and procedural issues might result in suboptimal plans for some patients that might go undetected. In this study, we prospectively investigated the clinical usefulness of an independent DVH prediction tool to detect outliers in treatment plan quality for prostate cancer patients. Material and Methods All prostate cancer patients treated from January 2015 till half September 2016 with the full prescribe ed dose delivered to the prostate only or to the prostate+seminal vesicles were included in the study. They were treated with an automatically generated VMAT or dMLC plan. The QA method was based on overlap volume histogram and principal component analysis and is fully independent of the planning method. The model was trained with 50% of the patients treated in 2015 (N=22) and validated on the other 50% (N=21). We focused on 5 different dose metrics: rectum Dmean, V65, V75; anus Dmean and bladder Dmean. Next, to study the clinical usefulness of treatment planning QA, the QA model was applied prospectively for the patients treated in 2016 (N=50). Patients for which at least one of the five dose metrics fell outside the 90% prediction confidence interval (CI) were further improved by manual plan adjustments (‘re-planning’). The replanning goals were to keep or improve all dose metrics of interest within or lower than the 90% CI, and anyway not
17 Patients from the prospective cohort were classified as outliers, including all four patients with metal hips, which were excluded from further analysis. The remaining outliers 13/46 (28.3%) were re-planned and for all the replanning requirements (above) were met. As shown in Figure 2, the new plans were moderately superior to the clinical plans for rectum Dmean (average improvement 0.9Gy, max. improvement 3.0Gy, p=0.009), V65 (2.4%, max 4.2%, p =0.001), anus Dmean (1.5Gy, max 6.8Gy, p=0.004), and bladder Dmean (1.7Gy, max 5.1Gy, p=0.001). The rectum V75 of the new plans was slightly higher than with the original plan (0.2 %, p=0.028). No significant differences were found in PTV conformity or the femoral head Dmax.
S129 ESTRO 36 _______________________________________________________________________________________________
Conclusion The proposed plan QA tool can detect outliers with an accuracy of 3-4Gy and 2%-3% (90% CI). Totally 13/46 (28%) of the automatically generated plans were outliers. Indeed, for all of them re-planning resulted in an improved plan. This emphasizes the need for treatment planning QA, also for automated treatment planning. For manual treatment planning, the percentage of outliers is expected to be higher and therefore treatment planning QA is even more important. OC-0255 Practical use of principal component analysis in radiotherapy planning D. Christophides1, A. Gilbert2, A.L. Appelt2, J. Fenwick3, J. Lilley4, D. Sebag-Montefiore2 1 Leeds CRUK Centre and Leeds Institute of Can cer and Pathology, University of Leeds, Leeds, United Kingdom 2 Leeds Institute of Cancer and Pathology - University of Leeds and Leeds Cancer Centre, St James’s University Hospital, Leeds, United Kingdom 3 Institute of Translational Medicine, University of Liverpool, Liverpool, United Kingdom 4 Leeds Cancer Centre, St James’s University Hospital, Leeds, United Kingdom Purpose or Objective Principal component analysis (PCA) is a promising technique for handling DVH data in NTCP modelling. However it is challenging to interpret its results clinically and use them to make informed decisions for specific patients. A method is developed that uses PCA-based NTCP modelling to produce treatment optimisation objectives which can be used for treatment plan improvement. The utility of the method is demonstrated in a treatment planning case as well as in a simulation study, for reducing predicted patient reported outcome (PRO) scores of vaginal stenosis. Material and Methods Data from 221 female patients treated with pelvic radiotherapy were made available from a larger study (DRF-2012-05-201) on optimising patient outcomes. Vaginal stenosis PRO scores (“Has your vagina felt tight?”: “Not at all” (0), “A little” (1), “Quite a bit” (2) and “Very much” (3)) were completed by 74 (29%) patients. The
principal components (PCs) extracted from the available external genitalia DVHs, along with clinical factors, were used to construct an ordinal logistic regression model that predicted the probability of patients having vaginal stenosis symptoms. The model identified age, hormone replacement therapy and the first PC (PC1) as important predictors of vaginal stenosis PRO scores. Based on the model, the probability of grade 2 or greater PRO score could be calculated; as well as a PC1 that could theoretically reduce that probability by 50% (PC1'). PC1' was used to derive a PCAmodified DVH' using the following method: i) the modified principal components were inversely transformed into the DVH domain to obtain a new DVH', ii) DVH' was cropped so the volumes were always greater than 0% and lower than the original DVH, and iii) DVH' was made monotonically decreasing. An anal cancer patient case was planned using VMAT and the PCA-based model information, as a demonstration of the clinical applicability of PCA-based modelling. The method was then used to modify the DVHs of all available patients (N=221). The probability of having grade 2 ≥ PRO scores using the un-modified patient DVH and the PCAmodified DVH' were compared using a paired t-test. Results The treatment planning case demonstrated the clinical relevance of PCA-based modelling by using PCA information to formulate cost functions to reduce the dose to the genitalia (Fig.1), which resulted in a reduction of the predicted probability of vaginal stenosis symptoms (Fig.2). The simulation results showed a statistically significant decrease in the probability of having grade 2 ≥ PRO scores (Reduction in mean = 33%, p<0.001).
deteriorate them by more than 1Gy/1% compared to the original plan. Given the 5 parameters of interest and the 90% criteria, (1- (0.9)5 )≈40% patients were expected to fall outside the prediction range. Results Figure 1 shows the results of the model validation.
Lung dataset size 16 46 61 69 74 HN Proton size
169 130 153 173 192
HN Photon size 68 136 145 121 136 Table 1: Plan datasets required for desired dose accuracy.
Figure 1 Comparison of clinical plan developed without (A) and with (B) predictive modeling. Conclusion We have demonstrated the ability to predict dosimetric indices. These results have clinical implications that extend from decision making to planning workflow improvement to quality improvement. OC-0254 Prospective validation of independent DVH prediction for QA of automatic treatment planning Y. Wang1, B.J.M. Heijmen1, S.F. Petit1 1 Erasmus MC - Cancer Institute, Radiation Oncology, Rotterdam, The Netherlands Purpose or Objective In our institute, fully automated, knowledge-based treatment planning is used in routine clinical practice. For the majority of patients, this is expected to result in high quality treatment plans. However, technical and procedural issues might result in suboptimal plans for some patients that might go undetected. In this study, we prospectively investigated the clinical usefulness of an independent DVH prediction tool to detect outliers in treatment plan quality for prostate cancer patients. Material and Methods All prostate cancer patients treated from January 2015 till half September 2016 with the full prescribe ed dose delivered to the prostate only or to the prostate+seminal vesicles were included in the study. They were treated with an automatically generated VMAT or dMLC plan. The QA method was based on overlap volume histogram and principal component analysis and is fully independent of the planning method. The model was trained with 50% of the patients treated in 2015 (N=22) and validated on the other 50% (N=21). We focused on 5 different dose metrics: rectum Dmean, V65, V75; anus Dmean and bladder Dmean. Next, to study the clinical usefulness of treatment planning QA, the QA model was applied prospectively for the patients treated in 2016 (N=50). Patients for which at least one of the five dose metrics fell outside the 90% prediction confidence interval (CI) were further improved by manual plan adjustments (‘re-planning’). The replanning goals were to keep or improve all dose metrics of interest within or lower than the 90% CI, and anyway not
17 Patients from the prospective cohort were classified as outliers, including all four patients with metal hips, which were excluded from further analysis. The remaining outliers 13/46 (28.3%) were re-planned and for all the replanning requirements (above) were met. As shown in Figure 2, the new plans were moderately superior to the clinical plans for rectum Dmean (average improvement 0.9Gy, max. improvement 3.0Gy, p=0.009), V65 (2.4%, max 4.2%, p =0.001), anus Dmean (1.5Gy, max 6.8Gy, p=0.004), and bladder Dmean (1.7Gy, max 5.1Gy, p=0.001). The rectum V75 of the new plans was slightly higher than with the original plan (0.2 %, p=0.028). No significant differences were found in PTV conformity or the femoral head Dmax.
S129 ESTRO 36 _______________________________________________________________________________________________
Conclusion The proposed plan QA tool can detect outliers with an accuracy of 3-4Gy and 2%-3% (90% CI). Totally 13/46 (28%) of the automatically generated plans were outliers. Indeed, for all of them re-planning resulted in an improved plan. This emphasizes the need for treatment planning QA, also for automated treatment planning. For manual treatment planning, the percentage of outliers is expected to be higher and therefore treatment planning QA is even more important. OC-0255 Practical use of principal component analysis in radiotherapy planning D. Christophides1, A. Gilbert2, A.L. Appelt2, J. Fenwick3, J. Lilley4, D. Sebag-Montefiore2 1 Leeds CRUK Centre and Leeds Institute of Can cer and Pathology, University of Leeds, Leeds, United Kingdom 2 Leeds Institute of Cancer and Pathology - University of Leeds and Leeds Cancer Centre, St James’s University Hospital, Leeds, United Kingdom 3 Institute of Translational Medicine, University of Liverpool, Liverpool, United Kingdom 4 Leeds Cancer Centre, St James’s University Hospital, Leeds, United Kingdom Purpose or Objective Principal component analysis (PCA) is a promising technique for handling DVH data in NTCP modelling. However it is challenging to interpret its results clinically and use them to make informed decisions for specific patients. A method is developed that uses PCA-based NTCP modelling to produce treatment optimisation objectives which can be used for treatment plan improvement. The utility of the method is demonstrated in a treatment planning case as well as in a simulation study, for reducing predicted patient reported outcome (PRO) scores of vaginal stenosis. Material and Methods Data from 221 female patients treated with pelvic radiotherapy were made available from a larger study (DRF-2012-05-201) on optimising patient outcomes. Vaginal stenosis PRO scores (“Has your vagina felt tight?”: “Not at all” (0), “A little” (1), “Quite a bit” (2) and “Very much” (3)) were completed by 74 (29%) patients. The
principal components (PCs) extracted from the available external genitalia DVHs, along with clinical factors, were used to construct an ordinal logistic regression model that predicted the probability of patients having vaginal stenosis symptoms. The model identified age, hormone replacement therapy and the first PC (PC1) as important predictors of vaginal stenosis PRO scores. Based on the model, the probability of grade 2 or greater PRO score could be calculated; as well as a PC1 that could theoretically reduce that probability by 50% (PC1'). PC1' was used to derive a PCAmodified DVH' using the following method: i) the modified principal components were inversely transformed into the DVH domain to obtain a new DVH', ii) DVH' was cropped so the volumes were always greater than 0% and lower than the original DVH, and iii) DVH' was made monotonically decreasing. An anal cancer patient case was planned using VMAT and the PCA-based model information, as a demonstration of the clinical applicability of PCA-based modelling. The method was then used to modify the DVHs of all available patients (N=221). The probability of having grade 2 ≥ PRO scores using the un-modified patient DVH and the PCAmodified DVH' were compared using a paired t-test. Results The treatment planning case demonstrated the clinical relevance of PCA-based modelling by using PCA information to formulate cost functions to reduce the dose to the genitalia (Fig.1), which resulted in a reduction of the predicted probability of vaginal stenosis symptoms (Fig.2). The simulation results showed a statistically significant decrease in the probability of having grade 2 ≥ PRO scores (Reduction in mean = 33%, p<0.001).