PV QA 3 - Poster Viewing Q&A 3
Purpose/Objective(s):The objective of this study is to develop planning tool to predict parotid and Rectum mean dose just before the IMRT planning and after the OAR and target delineation.
Materials/Methods:In order to develop a mean dose predicting tool we retrospectively studied 60 HN IMRT cases (115 parotid structures - 80 as a training data and 35 as test data) and 70 Pelvic cases (60 as training rectum data and 10 test data). The IMRT planning was done in a treatment planning system using DMPO optimization with 6MV photon beam and 7 to 9 beam directions by various physicists. Plans were executed on a linear accelerator during the year 2011 to 2016. Before extracting the data, plans of each selected patient was verified for optimal plan quality. We further tried to reduce mean dose of OAR without compromising PTV coverage and hot dose. After obtaining the optimal plan we created the database by extracting the total parotid, rectum volume (Vp and ,Vr), volume of parotid, rectum overlaped with PTV (Vp’ and Vr’), mean dose of parotid and rectum (DmeanP and DmeanR) and dose prescribed to the PTV ( Dpd). The rectum and parotid data was analyzed separately, we found good correlation between the fractions of parotid and rectum volume overlapping the PTV and the ratio of mean dose to prescribed dose to PTV. Thus the exponential regression was done and mathematical model was generated separately for predicting Dmean based upon the fractional OAR-PTV overlap and prescribed dose. The general regression equation used to fit the data was Dmean/Dpd = A x (B – Exp (-C x VOAR’ / VOAR), where A, B, C are regression parameters to best fit the curve and VOAR is either Vp or Vr. The rectum and parotid prediction model was validated separately by predicting the mean dose (Dpred) of 35 parotid and 10 rectum volumes under test data and compared them with the planned mean dose. This model was incorporated in planning process to prospectively predict the parotid and rectum mean dose before the IMRT planning and after the OAR and target delineation.
Results:The data analysis of fractional OAR-PTV overlap vs DmeanOAR/Dpd showed correlation of 94.7% (p value <0.0001) and 95.5%( p value <0.0001) respectively for parotid and rectum. The residual standard deviation in regression analysis was 0.0583 and 0.063 respectively for parotid and rectum data in exponential regression. The max deviation between Dplan and Dpred for the test data showed the deviation of 5-6 Gy absolute dose for both rectum and parotid.Incorporation of this model, helped oncologist to review their CTV in case of the predicted mean dose is higher than the tolerance level.
Conclusion:The developed prediction model helped the planner to decide the lower achievable level of parotid dose. This model also served oncologist to alter the overdrawn CTV without compromising treatment area, which reduces the overlap of parotid with PTV.It can be successfully implemented in the routine planning process to increase efficiency of the treatment planning.
The asset you are trying to access is locked. Please enter your access key to unlock.