Erschienen in:
16.04.2019 | Ultrasound
Gynecology Imaging Reporting and Data System (GI-RADS): diagnostic performance and inter-reviewer agreement
verfasst von:
Mohammad Abd Alkhalik Basha, Rania Refaat, Safaa A. Ibrahim, Nadia M. Madkour, Awad Mahmoud Awad, Elshaimaa Mohamed Mohamed, Ahmed A. El Sammak, Mohamed M. A. Zaitoun, Hitham A. Dawoud, Mai E. M. Khamis, Heba A. E. Mohamed, Ahmed Mohamed El-Maghraby, Ahmed A. El-Hamid M. Abdalla, Mostafa Mohamad Assy, Mohamad Gamal Nada, Ahmed Ali Obaya, Eman H. Abdelbary
Erschienen in:
European Radiology
|
Ausgabe 11/2019
Einloggen, um Zugang zu erhalten
Abstract
Objective
To evaluate diagnostic performance and inter-reviewer agreement (IRA) of the Gynecologic Imaging Reporting and Data System (GI-RADS) for diagnosis of adnexal masses (AMs) by pelvic ultrasound (US).
Patients and methods
A prospective multicenter study included 308 women (mean age, 41 ± 12.5 years; range, 15–73 years) with 325 AMs detected by US. All US examinations were analyzed, and AMs were categorized into five categories according to the GI-RADS classification. We used histopathology and US follow-up as the reference standards for calculating diagnostic performance of GI-RADS for detecting malignant AMs. The Fleiss kappa (κ) tests were applied to evaluate the IRA of GI-RADS scoring results for predicting malignant AMs.
Results
A total of 325 AMs were evaluated: 127 (39.1%) were malignant and 198 (60.9%) were benign. Of 95 AMs categorized as GI-RADS 2 (GR2), none was malignant; of 94 AMs categorized as GR3, three were malignant; of 13 AMs categorized as GR4, six were malignant; and of 123 AMs categorized as GR5, 118 were malignant. On a lesion-based analysis, the GI-RADS had a sensitivity, a specificity, and an accuracy of 92.9%, 97.5%, and 95.7%, respectively, when regarding only those AMs classified as GR5 for predicting malignancy. Considering combined GR4 and GR5 as a predictor for malignancy, the sensitivity, specificity, and accuracy of GI-RADS were 97.6%, 93.9%, and 95.4%, respectively. The IRA of the GI-RADS category was very good (κ = 0.896). The best cutoff value for predicting malignant AMs was >GR3.
Conclusions
The GI-RADS is very valuable for improving US structural reports.
Key Points
• There is still a lack of a standard in the assessment of AMs.
• GI-RADS is very valuable for improving US structural reports of AMs.
• GI-RADS criteria are easy and work at least as well as IOTA.