PURPOSE: To evaluate the accuracy of ultrasonographic (US) features of aden
omyosis by correlating them with histologic findings and to assess inter- a
nd intraobserver agreement.
MATERIALS AND METHODS: US was performed and videotaped in 102 consecutive h
ysterectomy specimens in a water bath. Videotapes were reviewed initially b
y two independent radiologists blinded to the clinical and histologic findi
ngs and after 1 month by one of the two; US and histologic findings were co
rrelated. Features evaluated included diffuse abnormal echotexture of myome
trium, :subendometrial myometrial cysts, subendometrial echogenic nodules,
subendometrial echogenic linear striations, nodular endometrial-myometrial
junction, poor definition of the endometrial-myometrial junction, asymmetri
c thickness of the anteroposterior wall of the myometrium, and globular con
figuration.
RESULTS: The prevalence of adenomyosis in this cohort was 29.4% (30 of 102
specimens). The mean sensitivity, specificity, negative predictive value, p
ositive predictive value (PPV), and accuracy for the diagnosis of adenomyos
is for the three reviews were 81%, 71%, 90%, 54%, and 74%, respectively. Al
l findings evaluated except for nodular endometrial-myometrial junction, we
re significantly more common in uteri with adenomyosis (P < .05). Heterogen
eous myometrium reached borderline significance (P = .05). The specificitie
s and PPVs of subendometrial striations, subendometrial echogenic nodules,
and asymmetric myometrial thickness were significantly higher than those of
other features (P < .05). The interobserver agreement was moderate (kappa
= 0.4), and the intraobserver agreement was good (kappa = 0.67) for the thr
ee reviews.
CONCLUSION: The presence of subendometrial linear striations, subendometria
l echogenic nodules, or asymmetric myometrial thickness improves the specif
icity and PPV of US in diagnosing adenomyosis.