Magnetic resonance imaging findings of cystic ovarian tumors: major differential diagnoses in five types frequently encountered in daily clinical practice

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

PURPOSE

To retrospectively evaluate the diagnostic potential of magnetic resonance imaging (MRI)-based features and radiomics analysis (RA)-based features for discriminating ovarian clear cell carcinoma (CCC) from endometrioid carcinoma (EC).

MATERIALS AND METHODS

Thirty-five patients with 40 ECs and 42 patients with 43 CCCs who underwent pretherapeutic MRI examinations between 2011 and 2022 were enrolled. MRI-based features of the two groups were compared. RA-based features were extracted from the whole tumor volume on T2-weighted images (T2WI), contrast-enhanced T1-weighted images (cT1WI), and apparent diffusion coefficient (ADC) maps. The least absolute shrinkage and selection operator (LASSO) regression with tenfold cross-validation method was performed to select features. Logistic regression analysis was conducted to construct the discriminating models. Receiver operating characteristic curve (ROC) analyses were performed to predict CCC.

RESULTS

Four features with the highest absolute value of the LASSO algorithm were selected for the MRI-based, RA-based, and combined models: the ADC value, absence of thickening of the uterine endometrium, absence of peritoneal dissemination, and growth pattern of the solid component for the MRI-based model; Gray-Level Run Length Matrix (GLRLM) Long Run Low Gray-Level Emphasis (LRLGLE) on T2WI, spherical disproportion and Gray-Level Size Zone Matrix (GLSZM), Large Zone High Gray-Level Emphasis (LZHGE) on cT1WI, and GLSZM Normalized Gray-Level Nonuniformity (NGLN) on ADC map for the RA-based model; and the ADC value, spherical disproportion and GLSZM_LZHGE on cT1WI, and GLSZM_NGLN on ADC map for the combined model. Area under the ROC curves of those models were 0.895, 0.910, and 0.956. The diagnostic performance of the combined model was significantly superior (p = 0.02) to that of the MRI-based model. No significant differences were observed between the combined and RA-based models.

CONCLUSION

Conventional MRI-based analysis can effectively distinguish CCC from EC. The combination of RA-based features with MRI-based features may assist in differentiating between the two diseases.

Atypical Endometriosis: A Comprehensive Systematic Review of Pathological Patterns and Diagnostic Challenges

Endometriosis is a benign condition affecting women of reproductive age. A potential association with ovarian cancer has been documented. Atypical endometriosis (AE) is characterized by deviations from the typical microscopic appearance of endometriosis, including cytologic and architectural atypia. AE has been recognized as a potential precursor to endometriosis-associated ovarian cancers (EAOC), particularly endometrioid and clear cell subtypes. AE presents challenges in diagnosis due to its diverse clinical and pathological features, often requiring careful histological evaluation for accurate identification. Architectural AE, defined by localized proliferation of crowded glands with atypical epithelium resembling endometrial neoplasia, and cytologic AE, characterized by nuclear atypia within the epithelial lining of endometriotic cysts, are key subtypes. Immunohistochemical and molecular studies have revealed aberrant expression of markers such as Ki67, COX-2, BAF250a, p53, estrogen receptor, progesterone receptor, and IMP-3. Long-term follow-up studies suggest relatively low recurrence and malignant transformation rates among patients with AE, but uncertainties persist regarding its exact malignancy potential and optimal management strategies. Integration of artificial intelligence and shared molecular aberrations between AE and EAOC may enhance diagnostic accuracy. Continuous interdisciplinary collaboration and ongoing research efforts are crucial for a deeper understanding of the relationship between endometriosis and carcinogenesis, ultimately improving patient care and surveillance.

An algorithm for the pre-operative differentiation of benign ovarian tumours based on magnetic resonance imaging interpretation in a regional core hospital: A retrospective study

Objective

For selecting minimally invasive surgery (i.e. laparoscopic ovarian cystectomy) for treating ovarian tumours (OTs) in premenopausal patients, the pre-operative differentiation of benign ovarian tumours (Be-OTs) based on magnetic resonance imaging (MRI) interpretation is important. This paper describes the authors' 8-year experience of approximately 1000 OT cases, and provides information about a diagnostic algorithm to help other hospitals.

Study design

The medical records of 901 patients aged < 50 years with OTs from 1 January 2015-31 March 31 2023 were reviewed. First, the accuracy of pre-operative differentiation between Be-OTs and borderline/malignant ovarian tumours (Bo/Ma-OTs) was compared in each type of OT. Second, to identify the factors influencing differentiation between Be-OTs and Bo/Ma-OTs in 164 serous/mucinous ovarian tumours (SM-OTs), a multi-variate logistic regression analysis was performed to assess the effect of 13 factors, including MRI findings, OT size and tumour markers.

Results

In the comparison of diagnostic accuracy of pre-operative MRI for each OT type, accuracy was found to be notably high for ovarian endometrial cyst (OEC) (n = 409), ovarian mature cystic teratoma (OMCT) (n = 308), ovarian endometrioid adenocarcinoma (OEA) (n = 6) and ovarian clear cell adenocarcinoma (OCCA) (n = 14). On the other hand, discrepancies between MRI and pathological findings often occurred in SM-OTs, including ovarian serous cystadenoma (n = 86), ovarian mucinous adenocarcinoma (n = 61), ovarian serous adenocarcinoma (n = 12) and ovarian mucinous adenocarcinoma (n = 5). In the multi-variate logistic regression analysis of the latter 164 patients, in addition to MRI findings, OT size and carbohydrate antigen 125 also had an effect to some extent. The combination of MRI interpretation and OT size may enhance differentiation of Be-OTs and Bo/Ma-OTs.

Conclusions

Among four types of OTs (OEC, OMCT, OEA and OCCA), MRI interpretation was able to differentiate between Be-OTs and Bo/Ma-OTs almost perfectly. Additionally, to mitigate the difficulty in differentiating SM-OTs, OT size may be useful in combination with MRI findings, although further accumulation and analysis of OT cases is needed.

Characterization of Primary Mucinous Ovarian Cancer by Diffusion-Weighted and Dynamic Contrast Enhancement MRI in Comparison with Serous Ovarian Cancer

(1) Background. The purpose of this study is to evaluate the diagnostic accuracy of a quantitative analysis of diffusion-weighted imaging (DWI) and dynamic contrast enhanced (DCE) MRI of mucinous ovarian cancer (MOC). It also aims to differentiate between low grade serous carcinoma (LGSC), high-grade serous carcinoma (HGSC) and MOC in primary tumors. (2) Materials and Methods. Sixty-six patients with histologically confirmed primary epithelial ovarian cancer (EOC) were included in the study. Patients were divided into three groups: MOC, LGSC and HGSC. In the preoperative DWI and DCE MRI, selected parameters were measured: apparent diffusion coefficients (ADC), time to peak (TTP), and perfusion maximum enhancement (Perf. Max. En.). ROI comprised a small circle placed in the solid part of the primary tumor. The Shapiro-Wilk test was used to test whether the variable had a normal distribution. The Kruskal-Wallis ANOVA test was used to determine the p-value needed to compare the median values of interval variables. (3) Results. The highest median ADC values were found in MOC, followed by LGSC, and the lowest in HGSC. All differences were statistically significant (p < 0.000001). This was also confirmed by the ROC curve analysis for MOC and HGSC, showing that ADC had excellent diagnostic accuracy in differentiating between MOC and HGSC (p < 0.001). In the type I EOCs, i.e., MOC and LGSC, ADC has less differential value (p = 0.032), and TTP can be considered the most valuable parameter for diagnostic accuracy (p < 0.001). (4) Conclusions. DWI and DCE appear to be very good diagnostic tools in differentiating between serous carcinomas (LGSC, HGSC) and MOC. Significant differences in median ADC values between MOC and LGSC compared with those between MOC and HGSC indicate the usefulness of DWI in differentiating between less and more aggressive types of EOC, not only among the most common serous carcinomas. ROC curve analysis showed that ADC had excellent diagnostic accuracy in differentiating between MOC and HGSC. In contrast, TTP showed the greatest value for differentiating between LGSC and MOC.