Program Information
The Tumor Segmentation in Rectal Cancer for Radiomics Analysis
J Wang*, P Hu , J Gan , L Shen , R Luo , Z Zhang , W Hu , Fudan University Shanghai Cancer Center, Shanghai, shanghai
Presentations
MO-RPM-GePD-J(A)-5 (Monday, July 31, 2017) 3:45 PM - 4:15 PM Room: Joint Imaging-Therapy ePoster Lounge - A
Purpose: Tumor delineation is critical for radiomics study, which have impact on subsequent step. In this study, we evaluated the reproducibility of the radiomics features for tumor segmentation in rectal cancer.
Methods: Thrity-one rectum patients’ CT and MR image were collected. Three tumor delineations were implemented by two physicians. One of the delineation was based on CT image only by physician A, two delineations were based on the CT and MR fusion images by physician A and B. After delineation, radiomics features were calculated on CT images. Three comparisons were implemented to analyze the reproducibility of the features, including the comparison between delineations with MR fusion and without MR fusion by physician A; the comparison between delineations with MR fusion by physician A and physician B; the comparison between the delineation without MR fusion by physician A and the delineation with MR fusion by physician B. Intraclass correlation coefficient was calculated in order to quantify the feature reproducibility.
Results: The reproducibility was improved by CT MR fusion. The number of features’ ICC larger than 0.8 for CT vs. CT-MR fusion was 8. The number of features’ ICC larger than 0.8 for CT vs. CT-MR fusion was 18. 197 features (74%) were improved by fusion. The ICC value from the comparison between delineation with MR fusion and without MR fusion showed that many features do not have high reproducibility even by one physician delineation.
Conclusion: The MR fusion will improve the reproducibility of radiomics features in rectal cancer.
Contact Email: