The Value of MRI Findings Combined With Texture Analysis in the Differential Diagnosis of Primary Ovarian Granulosa Cell Tumors and Ovarian Thecoma–Fibrothecoma

Abstract

ObjectiveThis study aims to explore the value of magnetic resonance imaging (MRI) and texture analysis (TA) in the differential diagnosis of ovarian granulosa cell tumors (OGCTs) and thecoma-fibrothecoma (OTCA–FTCA).MethodsThe preoperative MRI data of 32 patients with OTCA–FTCA and 14 patients with OGCTs, confirmed by pathological examination between June 2013 and August 2020, were retrospectively analyzed. The texture data of three-dimensional MRI scans based on T2-weighted imaging and clinical and conventional MRI features were analyzed and compared between tumor types. The Mann–Whitney U-test, χ2 test/Fisher exact test, and multivariate logistic regression analysis were used to identify differences between the OTCA–FTCA and OGCTs groups. A regression model was established by using binary logistic regression analysis, and receiver operating characteristic curve analysis was carried out to evaluate diagnostic efficiency.ResultsA multivariate analysis of the imaging-based features combined with TA revealed that intratumoral hemorrhage (OR = 0.037), log-sigma-20mm-3D_glszm_SmallAreaEmphasis (OR = 4.40), and log-sigma-2-0mm-3D_glszm_SmallAreaHighGrayLevelEmphasis (OR = 1.034) were independent features for discriminating between OGCTs and OTCA–FTCA (P < 0.05). An imaging-based diagnosis model, TA-based model, and combination model were established. The areas under the curve of the three models in predicting OGCTs and OTCA–FTCA were 0.935, 0.944, and 0.969, respectively; the sensitivities were 93.75, 93.75, and 96.87%, respectively; and the specificities were 85.71, 92.86, and 92.86%, respectively. The DeLong test indicated that the combination model had the highest predictive efficiency (P < 0.05), with no significant difference among the three models in differentiating between OGCTs and OTCA–FTCA (P > 0.05).ConclusionsCompared with OTCA–FTCA, intratumoral hemorrhage may be characteristic MR imaging features with OGCTs. Texture features can reflect the microheterogeneity of OGCTs and OTCA–FTCA. MRI signs and texture features can help differentiate between OGCTs and OTCA–FTCA and provide a more comprehensive and accurate basis for clinical treatment.

Diagnostic utility of a conventional MRI-based analysis and texture analysis for discriminating between ovarian thecoma-fibroma groups and ovarian granulosa cell tumors

“…The present results revealed that the diagnostic performance of the MRI-based and combination models was similar, and no additional benefit was obtained in TA. These results were contradictory to recent findings reported by Li et al showing that TA-based and combination models were superior to the MBF-based model [ 22 ]. However, there were important differences between that study and the present study; (1) The number of parameters they initially applied in TA was 1,316, which was too big for the small sample size of their study (46 cases).…”

“…Furthermore, it is time-consuming to delineate all slices. However, in a recent study, good findings were obtained for TA-based models using all slice data [ 22 ]; therefore, further studies are needed. The further development of our models will be achieved by addressing these issues in the future.…”

“…In pelvic MRI on ovarian sex cord-stromal tumors, recent radiomics studies reported the differentiation of OTFGs from uterine fibroids in the adnexal area [ 21 ]. A discriminative analysis of OGCTs and OTFGs using imaging-based models and TA-based models was conducted by Li et al [ 22 ]. Although there are clinical merits to predictive models for these adnexal lesions, the number of related studies is still limited, which may be due to the rarity of these diseases.…”

Establishment and validation of a radiological-radiomics model for predicting high-grade patterns of lung adenocarcinoma less than or equal to 3 cm

“…have also stated that inverse-variance is correlated with the survival time of glioblastoma patients ( 35 ). In another study, Li has reported SAHGLE as an independent feature to differentiate between primary ovarian granulosa cell tumor and ovarian thecoma-fibrothecoma (OR = 1.034) ( 36 ). Weng et al.…”

The feasibility of MRI texture analysis in distinguishing glioblastoma, anaplastic astrocytoma and anaplastic oligodendroglioma

“…The results suggested that machine learning-model had potential in improving accuracy of early diagnosis and reducing clinical workload. However, researches on highgrade gliomas were limited and most previous studies focused on distinguishing between two types of tumors (47)(48)(49)(50). For this reason, we employed radiomics-based machine learning technology to distinguish among GAM, AA and AO, and this multi-classification study was more suitable for clinical requirements.…”