This retrospective study, approved by the institutional review board and compliant with the Health Insurance Portability and Accountability Act of 1996, analyzed cases from April 2008 to February 2024. We searched our two Japanese hospital’s electronic medical record system for patients, and inclusion criteria were as follows: (1) surgically resected cases, (2) histologically diagnosed cases, and (3) patients who underwent preoperative MRI. The exclusion criteria were as follows: (1) histologically diagnosed ovarian metastasis from uterine endometrial cancer and (2) inappropriate image quality for evaluation. In total, 60 patients with histopathologically confirmed OEC were included. Patients ranged in age from 22 to 94 years, with a median age of 52. Of these, 20, 30, and 10 patients had grades 1, 2, and 3 OEC, respectively.

MRI was conducted using a 1.5-T (Intera Achieva 1.5-T Pulsar; Philips Healthcare, Best, The Netherlands), a 1.5-T (Inginea prodiva 1.5-T CS; Philips Healthcare, Best, The Netherlands), a 1.5-T (GENESIS SIGNA; GE Healthcare, The United States), a 3.0-T (Intera Achieva 3.0 T Quasar Dual; Philips Healthcare, Best, The Netherlands), or 3.0-T scanners (Inginea 3.0 T CX; Philips Healthcare, Best, The Netherlands). Forty-nine patients underwent unenhanced and contrast-enhanced MRI, while 11 patients had only unenhanced imaging. Imaging parameters included a section thickness of 4–7 mm with 1–2 mm intersection gap and a 24 × 24–44 × 44-cm field of view. Acquisitions included axial and sagittal T2-weighted fast spin-echo (TR/TE, 2,700–6423/81–130 ms), axial T1-weighted spin-echo (TR/TE, 367–808/7–11 ms), and axial fat-suppressed T1-weighted fast spin-echo (TR/TE, 367–808/7–10 ms) sequences. In addition, axial diffusion-weighted single shot spin-echo echo-planar (TR/TE, 2,700–5,000/69–80 ms; b-value = 0 and 1000 s/mm2) was performed in 52 patients, axial and coronal or sagittal fat-suppressed gadolinium-enhanced T1-weighted spin-echo (TR/TE, 367–816/6.7–12 ms) imaging was obtained in 49 patients following the intravenous injection of 0.1 mmol/kg gadopentetate dimeglumine (Magnevist, Bayer HealthCare, Leverkusen, Germany) or gadobutrol (Gadavist, Bayer HealthCare, Leverkusen, Germany).

Two radiologists, with 25- and 11-year of experience in gynecological imaging, independently evaluated all MRI scans. Discrepancies were resolved by consensus. Reviewers were blinded to clinical and pathological data.

First, the reviewers conducted a qualitative assessment of each case by evaluating several parameters, including the margin (well-defined or ill-defined), configuration (pure solid, mixed solid and cystic, or pure cystic), and predominance (cystic or solid). If the EOCs involved bilateral ovaries, the reviewer evaluated the imaging findings of larger lesion. In addition, they assessed the presence of uterine corpus cancer, uterine adenomyosis, pelvic endometriosis, lymphadenopathy, abnormal ascites, and peritoneal dissemination. Lymphadenopathy was defined as a short-axis diameter > 8 mm within the pelvis. Peritoneal dissemination was identified by nodular or smooth thickening of the peritoneum. Abnormal ascites was defined as ascites exceeding the uterine fundus and/or filling the pelvic cavity [17, 18].

Second, the reviewers evaluated the qualitative imaging findings of the cystic component. Parameters included loculation (unilocular or multilocular), signal intensity on T1- (low, iso-, or high) and T2- (iso-, mildly high or high) weighted images, T1-hyperintense cyst, and fluid–fluid levels were evaluated. The signal intensity of the cystic component was evaluated at the cyst with the largest solid component and was compared with that of the iliopsoas muscle. If the lesion had multiple cysts within the tumors, T1-hyperintense cyst was positive when at least one hyperintense cyst relative to the iliopsoas muscle was observed on T1-weighted images.

Third, the qualitative imaging findings for solid components were evaluated based on exophytic growth, mural nodule growth pattern (eccentric or centripetal), surface lobulation, signal intensity (iso-, mildly high, or high), and homogeneity (homogeneous or heterogeneous) on T1- and T2-weighted images. Eccentric and centripetal pattern were defined as numerous solid components existed along the inner cystic surface and a few solid proportions in the cystic mass, respectively [8]. Reviewers noted T2-hyperintense and T2-hypointense areas within the solid components. Signal intensity of solid components was compared with the iliopsoas muscle on T1- and T2-weighted images. If multiple solid components were present, the largest solid component was analyzed. T2-hyperintense or T2-hypointense solid component was present if any hyperintense or hypointense area was observed within the solid component on T2-weighted images. On contrast-enhanced T1-weighted images, enhancement degree (marked or mild), homogeneity (homogeneous or heterogeneous), and necrosis were recorded. Necrosis was defined as ill-defined, unenhanced regions within solid components.

Finally, the quantitative imaging findings included the maximum diameter of the whole tumor and solid component, the height of the solid component, the signal intensity ratio of cystic and solid components on T1- and T2-weighted images, and the signal intensity ratio of solid component on contrast-enhanced T1-weighted images. The apparent diffusion coefficient (ADC) value of the solid component was also measured. Regions of interest (ROIs) were placed on both the cystic and solid components as well as the iliopsoas muscle to calculate signal intensity ratios on T1-, T2-, and contrast-enhanced T1-weighted images. Solid components on contrast-enhanced T1-weighted images were considered as enhanced areas by referring to unenhanced T1-weighted images. The ratio of cystic or solid components to the muscle signal intensity was calculated. Signal intensity of cystic and solid component was measured by placing ROIs on the cyst with the largest solid component and the largest solid component, respectively. The ADC value was determined using ADC maps by placing ROIs within the largest solid component, excluding necrotic and cystic areas based on T2- and contrast-enhanced T1-weighted images [18].

All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan) [19]. Quantitative results of grade 1 and grade 2–3 OECs were compared using the Mann–Whitney U test and those of three grades were compared using the Kruskal–Wallis test, while qualitative results were analyzed using Fisher’s exact test. Receiver operation characteristics (ROC) curve analysis was used to determine the performance of quantitative imaging findings, and area under the curve (AUC) was calculated to establish the optimal cutoff value for differentiating grade 1 OEC from grade 2–3 OECs. The optimal ROC cutoff value was determined based on the Youden criterion. Moreover, the sensitivity, specificity, and accuracy for differentiating grade 1 OEC from grade 2–3 OECs were calculated. A p value < 0.05 was considered statistically significant. Inter-observer variability for qualitative assessments was measured using kappa statistics.