Crohn's Disease Evaluated with Magnetic Resonance Enteroclysis: Diagnostic Performance of Experienced and Inexperienced Readers before and after Training

Background: Magnetic resonance enteroclysis (MRE) is suggested to become the preferred radiological method in small-bowel Crohn's disease (CD). However, the performance of inexperienced readers may influence the diagnostic value of the method and has not been previously investigated. Purpose: To compare readings of MRE in small-bowel CD performed by experienced and inexperienced readers before and after training.Material and Methods: One experienced radiologist (observer 1) and two trainees (observers 2 and 3) reviewed 60 MRE examinations. A second reading was performed after training. Bowel wall thickness (BWT), ulcers (BWU), stenosis (BWS), fistulas (FIS), and abscesses (ABS) were evaluated. A reference standard based on clinical records was established. Results: BWT in the terminal ileum was evaluated with high diagnostic performance (sensitivity: observer 1, 83%; observer 2, 72%; observer 3, 78%). Only BWU was diagnosed with a higher sensitivity by observer 1 (78% vs. 33% and 39%, respectively; P = 0.02). False-positive findings for BWT in the jejunum (observer 2: 7; observer 3: 4) and fistulas and abscesses (observer 2: 11/5; observer 3: 5/4) were made by the trainees. Interobserver agreement in the jejunum was poor (observer 1/observer 2: kappa = 0.23; observer 1/observer 3: kappa = -0.03) and in the ileum good (observer 1/observer 2: kappa = 0.78; observer 1/observer 3: kappa = 0.73). After training, evaluation of BWU (observer 2: 56%, P = 0.22; observer 3: 44%, P = 0.03), BWT (observer 2: 2; observer 3: 2), and interobserver agreement in the jejunum improved (observer 1/observer 2: kappa = 0.66; observer 1/observer 3: kappa = 0.66). However, the number of diagnosed fistulas and abscesses remained high. Conclusion: Before training, most findings of Crohn's disease in the terminal ileum were evaluated with high diagnostic performance by all readers. However, the inexperienced readers evaluated BWU with a low sensitivity and overestimated the number of FIS, number of ABS, and increased BWT in the jejunum. After training, evaluation by inexperienced readers of BWU and increased BWT in the jejunum improved.