MRI criteria for adenomyosis: interobserver reproducibility of diagnostic features

Diagnostic accuracy public-domain-us
View on PubMed View at publisher

Abstract

PURPOSE: To determine which MRI features used in the diagnosis of adenomyosis are reproducible between readers, and whether a diagnostic certainty scale used in routine reporting can be applied consistently. METHODS: Prospective single-center study of 108 consecutive women referred for 3-T pelvic MRI after transvaginal ultrasound suspicion of adenomyosis. Three subspecialized radiologists, with 10, 8 and 5 years of experience, independently assessed all examinations, blinded to clinical data, ultrasound findings and to each other. Diagnostic certainty was recorded on a four-level ordinal scale (definite, probable, possible, absent). Agreement was quantified with Fleiss' kappa for the three readers, Cohen's kappa for pairs, and weighted kappa for ordinal variables. RESULTS: Agreement for the four-level scale was fair when unweighted (κ = 0.401) and substantial when quadratic-weighted (mean 0.663). No patient was rated probable by all three readers, and agreement across the definite-versus-probable boundary was null (κ = -0.013 to 0.143); collapsing the two categories raised it to κ = 0.581-0.698. For the two direct criteria it was substantial for myometrial hemorrhagic foci (κ = 0.650) and moderate for myometrial cysts (κ = 0.555). Among junctional zone measurements, agreement was moderate for thickness > 12 mm (κ = 0.530) but absent for junctional zone irregularity (κ = 0.048), which was non-evaluable in 21 of 108 examinations (19.4%). These examinations showed a higher prevalence of motion artifacts (p = 0.0005). Agreement on severity grade was moderate (κ = 0.441). CONCLUSION: Reproducibility varied widely across MRI features of adenomyosis, and the intermediate levels of the diagnostic certainty scale were not applied consistently. These findings argue for redefining the current MRI criteria for adenomyosis.

My notes (saved in your browser only)

Citation neighborhood (no data yet)

We don't have any in-corpus citations linked to this paper yet. The paper's references may be in our DB but unresolved to ``paper_id`` (resolution happens at ingest when the cited DOI matches a row we already have). Run the cross-source citation reconcile pass to retry.

Source provenance

pubmed
last seen: 2026-11-17T06:04:02.938873+00:00
License: public-domain-us · commercial use OK · attribution required
Courtesy of the U.S. National Library of Medicine