Evaluation of interobserver variability and diagnostic performance of developed MRI-based radiological scoring system for invasive placenta previa

• Published in: Journal of magnetic resonance imaging 2016-09, Vol.44 (3), p.573-583
• Main Authors: Ueno, Yoshiko, Maeda, Tetsuo, Tanaka, Utaru, Tanimura, Kenji, Kitajima, Kazuhiro, Suenaga, Yuko, Takahashi, Satoru, Yamada, Hideto, Sugimura, Kazuro
• Format: Article
• Language: English
• Subjects: Adult; Algorithms; Clinical Competence; Female; Humans; Image Enhancement - methods; Image Interpretation, Computer-Assisted - methods; invasive placenta; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; MRI; NMR; Nuclear magnetic resonance; Observer Variation; Placenta; placenta accreta; placenta previa; Placenta Previa - diagnostic imaging; Placenta Previa - pathology; Pregnancy; Pregnancy complications; prenatal diagnosis; Reproducibility of Results; Sensitivity and Specificity; Severity of Illness Index; Small mammals
• ISSN: 1053-1807; 1522-2586
• DOI: 10.1002/jmri.25184

Purpose To evaluate the interobserver variability and diagnostic performance of a developed magnetic resonance imaging (MRI)‐based scoring system for invasive placenta previa. Materials and Methods Prenatal MR images of 70 women were retrospectively evaluated, 18 of whom were diagnosed with invasive placenta. The six MR features (dark band on T2‐weighted images, intraplacental abnormal vascularity, placental bulge, heterogeneous placenta, myometrial thinning, and placental protrusion sign) were scored on 5‐point Likert scale separately, and the cumulative radiological score (CRS) was defined as the sum of each score. Two more experienced radiologists (readers A and B) and two less experienced residents (readers C and D) calculated the CRS. Interobserver variability was assessed by measuring the intraclass correlation coefficient. Diagnostic performance was evaluated by means of receiver operating characteristic (ROC) analysis. Results Interobserver variability for CRS was excellent for the more experienced radiologists (0.85), and good for all readers (0.72) and the less experienced residents (0.66). The area under the ROC curve (Az) and accuracy (Acc) for CRS were significantly higher or equivalent to those of other MR features for all readers (Az and Acc for reader A; CRS, 0.92, 91.4%; intraplacental T2 dark band, 0.83, P = 0.009, 81.4%, P = 0.03; intraplacental abnormal vascularity, 0.9, P = 0.3, 90.0%, P = 1.00; placental bulge, 0.81, P = 0.0008, 80.0%, P = 0.02; heterogeneous placenta, 0.85, P = 0.11, 74.3%, P = 0.002; myometrial thinning, 0.84, P = 0.06, 60.0%, P < 0.0001; placental protrusion sign, 0.81, P = 0.01, 81.4%, P = 0.26). Conclusion This developed MRI‐based scoring system demonstrated excellent or good interobserver variability, and good diagnostic performance for invasive placenta previa. J. Magn. Reson. Imaging 2016;44:573–583.