Comparison of Dosiomics Features and Dosimetric Parameters for Detecting Variations in Dose Distribution in Breast Cancer Radiotherapy

Abstract

Background:This study aimed to investigate and compare the ability of dosiomics features (DFs) and dosimetric parameters (DPs) in detecting variations in dose distribution.Materials and Methods:This research included 15 patients with early-stage breast cancer who had previously undergone radiotherapy using three-dimensional (3D) conformal radiotherapy technique. Four treatment plans are created by different users for each patient. DPs such as D1(), D5(), and & mldr;, up to D100() were analyzed for each region of interest, and DFs were extracted from each plan using 3D-Slicer software. The coefficient of variation (CV) was used to measure the ability of each DFs or DPs to identify differences in dose distribution. CVs were calculated for intrapatient (across four plans) and interpatient (across one plan for all patients) comparisons.Results:Results showed that the planning target volume (PTV) and heart had the highest CV values in the gray level size zone matrix group (1.05, 0.68). The PTV showed the highest CV for SZM-large area low gray level emphasis, the lung for SZM-Small area low gray level emphasis, and the heart for SZM-size zone nonuniformity. For the D20() parameter, the heart had the highest CV, followed by the lung and PTV, with CVs of 0.7, 0.56, and 0.51, respectively.Conclusion:The findings suggest that DFs are more effective than DPs in differentiating between dose distributions. These features could play a key role in future radiotherapy plan evaluations with further study.