Magnetic resonance imaging-based radiomics analysis of the differential diagnosis of ovarian clear cell carcinoma and endometrioid carcinoma: a retrospective study

• DOI: 10.1007/s11604-024-01545-z
• Publication Date: 2024-03-12T19:25:36Z
• Authors: Nobuyuki Takeyama; Yasushi Sasaki; Yasuo Ueda; Yuki Tashiro; Eliko Tanaka; Kyoko Nagai; Miki Morioka; Takafumi Ogawa; Genshu Tate; Toshi Hashimoto; Yoshimitsu Ohgiya
• Keywords: Gray level

Abstract Purpose To retrospectively evaluate the diagnostic potential of magnetic resonance imaging (MRI)-based features and radiomics analysis (RA)-based for discriminating ovarian clear cell carcinoma (CCC) from endometrioid (EC). Materials methods Thirty-five patients with 40 ECs 42 43 CCCs who underwent pretherapeutic MRI examinations between 2011 2022 were enrolled. MRI-based two groups compared. RA-based extracted whole tumor volume on T2-weighted images (T2WI), contrast-enhanced T1-weighted (cT1WI), apparent diffusion coefficient (ADC) maps. The least absolute shrinkage selection operator (LASSO) regression tenfold cross-validation method was performed to select features. Logistic conducted construct models. Receiver operating characteristic curve (ROC) analyses predict CCC. Results Four highest value LASSO algorithm selected MRI-based, RA-based, combined models: ADC value, absence thickening uterine endometrium, peritoneal dissemination, growth pattern solid component model; Gray-Level Run Length Matrix (GLRLM) Long Low Emphasis (LRLGLE) T2WI, spherical disproportion Size Zone (GLSZM), Large High (LZHGE) cT1WI, GLSZM Normalized Nonuniformity (NGLN) map GLSZM_LZHGE GLSZM_NGLN model. Area under ROC curves those models 0.895, 0.910, 0.956. performance model significantly superior ( p = 0.02) that No significant differences observed Conclusion Conventional can effectively distinguish CCC EC. combination may assist in differentiating diseases.