Click here to


Are you sure ?

Yes, do it No, cancel

Novel Knowledge-Based Treatment Planning Model for Prostate Cancer Patients Treated with Hypofractionation and Simultaneous Integrated Boost

A Chatterjee*, M Serban , S Faria , L Souhami , F Cury , J Seuntjens , McGill University Health Centre, Montreal, QC


(Wednesday, 8/1/2018) 10:15 AM - 12:15 PM

Room: Davidson Ballroom A

Purpose: Treatment planning complexity increases with the number of planning target volumes (PTVs). We aimed to create a knowledge-based planning model for hypofractionated prostate cancer radiotherapy (with simultaneous integrated boost), using RapidPlanâ„¢ to establish best clinical planning practice and improve treatment planning efficiency.

Methods: RapidPlanâ„¢ creates a linear regression relationship between geometric features (OAR volumes, OAR-PTV distance, etc.) and corresponding dosimetric information. An initial model was trained using 48 patients treated with 60 Gy to prostate (PTV60) and 44 Gy to pelvic nodes (PTV44) in 20 fractions. To improve the goodness-of-fit of the model, an intermediate model was generated using the dose-volume histograms of best-spared OARs of the initial model. Using the intermediate model and further manual tweaking, all 48 cases were re-planned, and subsequently used to train the final model. The final model was validated on an independent set of 50 additional patients. All clinical plans used volumetric modulated arc therapy, and met clinical dose constraints of our institution.

Results: For both PTVs, V95% was similar (~99.9%) between clinical and knowledge-based plans. D1% was lower for model plans: by 1.2±0.1Gy (PTV60), and by 2.4±0.4Gy (PTV44). Model plans were more homogeneous, as evaluated using the homogeneity index (HI = (D1%-D99%)/D50%): 0.086 vs 0.064 (clinical vs model) for PTV60 (ΔHI = 0.022±0.002), and 0.158 vs 0.110 for PTV44 (ΔHI = 0.048±0.008). OAR sparing was superior for knowledge-based planning: ΔDmean = 3.7±0.4Gy (bladder), and 3.2±0.4Gy (rectum); ΔD2% = 1.2±0.3Gy (bowel bag), and 4.8±0.4Gy (femoral heads). All uncertainties are standard error of the mean. All stated improvements have p<0.001. Total knowledge-based planning time (typically under 30 minutes) was shorter than manual planning (typically 2.5 hours), and was limited by hardware, not by planner intervention.

Conclusion: Our knowledge-based planning model delivers efficient, consistent plans with excellent PTV coverage and improved OAR sparing.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email