Reproducibility of MRI Diagnosis of Female Genital Anomalies.

Purpose: To assess and compare intrareader and interreader reproducibility of magnetic resonance imaging (MRI) diagnosis of female genital anomalies (FGAs) using the American Society for Reproductive Medicine-Mullerian anomalies classification 2021 (ASRM-MAC 2021) and European Society of Human Reproduction and Embryology-European Society for Gynecological Endoscopy (ESHRE-ESGE) 2016 classification. Methods: In this retrospective study, we searched our electronic MRI database from April 2021 to September 2023, selecting MRI studies with FGAs. Seventy-six consecutive studies were included and reviewed by 4 independent radiologists using both classifications. Studies were re-evaluated after 1 month. Reproducibility was assessed using kappa (κ) scores with 95% confidence intervals (CI). Results: Intrareader agreement for MRI diagnosis of FGAs was substantial to excellent, with κ scores ranging from 0.684 (95% CI, 0.534-0.834) to 0.985 (95% CI, 0.963-1.01) using the ASRM-MAC 2021 and from 0.743 (95% CI, 0.621-0.865) to 0.846 (95% CI, 0.719-0.973) using the ESHRE-ESGE 2016 classification. Pairwise interreader agreement was higher with the ASRM-MAC 2021, ranging from moderate (κ = 0.491; 95% CI, 0.341-0.642) to substantial (κ = 0.709; 95% CI, 0.597-0.821), compared to the ESHRE-ESGE 2016 classification, with weak (κ = 0.080; 95% CI, 0.068-0.228) to moderate (κ = 0.511; 95% CI, 0.344-0.678) agreement. Overall interreader agreement was moderate for both classifications (κ = 0.599; 95% CI, 0.562-0.638 for ASRM-MAC 2021 and κ = 0.429; 95% CI, 0.396-0.463 for ESHRE-ESGE 2016 classification), but with significant differences (non-overlapping CIs). Conclusion: The intrareader reproducibility was high for both classifications, whereas the interreader reproducibility was higher using the ASRM-MAC 2021, highlighting the impact of classification criteria on the reproducibility of MRI diagnosis of FGAs.

Impact of Adding Mean Apparent Diffusion Coefficient (ADCmean) Measurements to O-RADS MRI Scoring For Adnexal Lesions Characterization: A Combined O-RADS MRI/ADCmean Approach.

RATIONALE AND OBJECTIVES: Evaluate the impact of adding mean apparent diffusion coefficient (ADCmean) measurements to the Ovarian-Adnexal Imaging Reporting and Data System MRI (O-RADS MRI) scoring for adnexal lesion characterization using a combined O-RADS MRI/ADCmean reading approach. MATERIALS AND METHODS: This prospective study included 90 women who underwent pelvic MRI for adnexal lesions diagnosis and characterization. Two readers scored the adnexal lesions using the O-RADS MRI scoring independently and in consensus. A third reader calculated ADCmean measurements. The final diagnoses were determined by histo-pathology (n = 77) or follow-up imaging (n = 13). Areas under the curves (AUCs) and diagnostic performance metrics were calculated for the O-RADS MRI scoring, ADCmean, and combined O-RADS MRI/ADCmean thresholds. P-value <0.05 was significant. RESULTS: 116 adnexal lesions (71 benign, 45 malignant) were analyzed. The optimal thresholds to predict malignant adnexal lesions were O-RADS MRI score >3 and ADCmean value ≤1.08 × 10-3 mm2/s (AUC 0.926 and 0.823; sensitivity 97.7% and 95.5%; specificity 87.3% and 68%; positive predictive value (PPV) 83% and 66.2%; positive likelihood ratio (PLR) 7.7 and 3.08, respectively). Compared to the O-RADS MRI scoring, a combined threshold of O-RADS MRI >3/ADCmean ≤1.08 × 10-3 mm2/s, yielded a reduction of false positives, a significant increase in the specificity (97.1%, p = 0.005), PPV (95.4%, p = 0.002), and PLR (33.1, p <0.0001), and non-significant change in the AUC (0.953, p = 0.252), and sensitivity (93.3%, p = 0.467). CONCLUSION: The diagnostic performance of O-RADS MRI scoring to characterize adnexal lesions could be improved by adding the ADCmean values through reducing false positives, increasing specificity, and maintaining good sensitivity.

Diagnostic accuracy and discriminative power of biparametric versus multiparametric MRI in predicting muscle-invasive bladder cancer.

PURPOSE: Investigate and compare the diagnostic accuracy and discriminative power of biparametric MRI (bp-MRI) and multiparametric MRI (mp-MRI) in predicting muscle-invasive bladder cancer (MIBC) based on Vesical Imaging-Reporting and Data System (VI-RADS) scoring and evaluate potentially influencing factors on both protocols' accuracy. METHOD: This retrospective study included 54 bladder cancer (BC) patients who underwent bladder MRI and histo-pathological assessment. Three readers independently reviewed the MRI studies and assigned a 1-5 score for T2-weighted, diffusion-weighted, and dynamic contrast-enhanced images. Then, bp-MRI and mp-MRI final VI-RADS scores were recorded for each BC. Diagnostic tables, chi-square test, kappa score (k), logistic regression, receiver operating characteristics (ROC) curves, areas under the curves (AUCs), and VI-RADS cut-off values were calculated. A Delong test was performed for ROC curve comparison. A P-value<0.05 was considered significant. RESULTS: In predicting MIBC, bp-MRI and mp-MRI had comparable diagnostic accuracy with insignificant differences for the three readers (P = 0.364,0.718,0.702). Radiologists' experience, and tumors' size and morphology had insignificant effect on bp-MRI accuracy (P = 0.086, 0.392,0.294), respectively. Tumors' size significantly influenced mp-MRI accuracy (P = 0.039). Bp-MRI and mp-MRI had comparable discriminative power with insignificant differences for all readers (P > 0.05). Using VI-RADS > 3 cut-off value improved the discriminative power of bp-MRI. Excellent inter-reader agreement in VI-RADS scoring for bp-MRI (k range, 0.814-0.867) and mp-MRI (k range, 0.787-0.859) was observed. CONCLUSION: Bp-MRI and mp-MRI demonstrated comparable diagnostic accuracy and discriminative power in predicting MIBC. The accuracy of bp-MRI was not influenced by radiologists' experience, or tumors' size and morphology.