Magnetic resonance texture analysis for predicting histologic tumor type in endometrial cancer.

e17580 Background: The aim of the present study was to assess the feasibility of texture analysis and histogram on magnetic resonance image for predicting histologic type in patients with endometrial cancer. Methods: A total of 85 patients who underwent surgery and had pathologically confirmed endometrial cancer were included in this study. We investigated the parameters for normalized perfusion map including perfusion ratio and integrated density, defined as the product of area and mean gray value from the normalized perfusion mapping on gadolinum-enhanced T1-weigted pelvic MRI. Apparent Diffusion Coefficient map and texture analysis by gray-level co-occurrence matrix in endometrial cancer, including angular second moment, contrast, correlation, entropy and inverse difference moment. MR parameters were correlated with pathologic feature including tumor type, FIGO staging, tumor invasion and lymph node metastasis. Results: The normalize perfusion map analysis revealed noteworthy distinctions between type 1 and type 2 endometrial cancer. Specifically, we observed significant differences in tumor area (p = 0.014), perfusion ratio (p = 0.001), perimeter (p = 0.002), integrated density (p = 0.042), median (p = 0.001), and roundness (p = 0.018). Entropy form the texture analysis demonstrated associations with integrated density and linear correlation, with notable differences observed between type 1 (R 2 = 0.363) and type 2 (R 2 = 0.471) endometrial cancer subtypes, displaying statistical significance (p = 0.010). Conclusions: In conclusion, parameters obtained by texture analysis and perfusion map could be useful for predicting the histological type of endometrial cancer.