Premium
SU‐E‐T‐341: Use of Patient Geometry and Multiple Linear Regression to Predict Prescription Dose for Pleurectomy/decortication Cases in Malignant Pleural Mesothelioma
Author(s) -
Dumane V,
Yuan Y,
Rimner A,
Yorke E,
Rosenzweig K
Publication year - 2014
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4888674
Subject(s) - medicine , nuclear medicine , lung , linear regression , lung volumes , radiation treatment planning , linear correlation , correlation , radiology , radiation therapy , mathematics , geometry , statistics
Purpose: To identify patient geometric parameters that correlate with the prescription dose in the treatment of malignant pleural mesothelioma (MPM) so as to build a model that could assist in the prediction of the same apriori and this help in planning. Methods: Planning CT scans for ten right sided and ten left sided patients were examined to extract five parameters for each case, which involved ratio of the ipsilateral lung volume to the total lung volume (X1), volume of PTV overlap with ipsilateral lung to ipsilateral lung volume (X2), volume of the PTV overlap with ipsilateral lung to total lung volume (X3), volume of the PTV overlap with ipsilateral lung to PTV volume (X4) and ratio of contralateral lung volume to total lung volume (X5). Each patient for this study was planned with VMAT using 6 MV on Eclipse V11 to a prescription dose that kept the mean total lung dose just under 20 Gy. Correlation coefficients were obtained between the parameters and the dose prescribed. Those that significantly correlated were combined using multiple linear regression. The model built was tested on 9 new cases. Results: All geometric parameters investigated except X2 showed significant correlation with the prescription dose. The correlation coefficients were r = −0.5421 for X1 (p = 0.0135), r = −0.6891 for X3 (p = 0.0008), r = −0.6797 for X4 (p = 0.001) and r = 0.5421 for X5 (p = 0.0135). The predictive model D = 4192.2×1 – 3577.1×3 – 471.6×4 + 6253.4×5 had a higher correlation (r = 0.86, p = 1.0657e‐06) with dose. For 8 out of the 9 test cases, the predicted dose was either equal to or upto 1 fraction of the prescription dose. Conclusion: The multiple regression model developed here could serve as a guide for dose prescribed in treatment of MPM.