Interobserver agreement and diagnostic performance of individual MRI criteria for diagnosis of placental adhesion disorders

Abstract

Aim: To assess the diagnostic accuracy of several criteria for the diagnosis of placental adhesion disorder (PAD) on magnetic resonance imaging (MRI) and to assess interobserver agreement and reader accuracy based on years of interpretive experience.

Materials and methods: Blinded evaluation of the placental MRI studies of 28 randomly selected women, 18 with PADs and 10 without PADs, was undertaken by three radiologists with 10, 5, and 2 years' experience in placental MRI interpretation. The presence of placenta praevia, dark intraplacental bands, heterogeneity, uterine bulging, "shaggy dog" appearance of the uterine serosa, subjective impression of extraplacental invasion, and dark intraplacental bands on diffusion-weighted imaging were assessed. Placental histology was reviewed blinded to the original reports and to MRI interpretation and this, along with clinical information at the time of delivery, formed the reference standard.

Results: Dark intraplacental bands on T2-weighted imaging were the most sensitive and specific as well as the most agreed upon (kappa=0.7) criterion for PAD for the three readers. Assessment of uterine bulging (kappa=0.42) and placental heterogeneity (kappa=0.48) did not improve diagnostic accuracy. DWI improved sensitivity but decreased specificity for the least experienced reader and did not change sensitivity or specificity for the more experienced readers.

Conclusion: Assessment of the placenta for dark bands is the most sensitive, specific, and reproducible criterion for diagnosis of PADs using MRI.