Intra- and interobserver variability in the diagnosis of GERD by real-time MRI

Abstract

The purpose of this study was to assess the reproducibility of functional and anatomical parameters of swallowing events as determined by real-time MRI at 40 ms temporal resolution (25 frames per second). Twenty-three consecutive patients with gastroesophageal reflux disease (GERD) underwent real-time MRI of the gastroesophageal junction at 3.0 T. Real-time MRI was based on highly undersampled radial fast low angle shot (FLASH) acquisitions with iterative image reconstruction by regularized nonlinear inversion (NLINV). MRI movies visualized the esophageal transport of a pineapple juice bolus, its passage through the gastroesophageal junction and functional responses during a Valsalva maneuver. His-angle, sphincter position, sphincter length and sphincter transit time were assessed by two radiologists. Interobserver and intraobserver intraclass correlation coefficients (ICC) were evaluated and Bland-Altman plots were constructed to assess the observer agreement. Interobserver agreement was excellent for sphincter transit time (ICC = 0.92), His-angle (ICC = 0.93), His-angle during Valsalva maneuver (ICC = 0.91) and sphincter-to-diaphragm distance (ICC = 0.98). Sphincter length and oesophageal diameter showed good interobserver agreement (ICC = 0.62 and ICC = 0.70). Intraobserver agreement was good for sphincter length (ICC = 0.80) and excellent for sphincter transit time, His-angle and His-angle during Valsalva maneuver, sphincter-to-diaphragm distance, and esophageal diameter (ICC = 0.91; ICC = 0.97; ICC = 0.97; ICC = 0.998; ICC = 0.93). All functional parameters of the gastroesophageal junction had good to excellent reproducibility. Visual assessment of Bland Altman plots did not reveal any systematic interobserver bias. In conclusion, the visualization of swallowing events by real-time MRI has a high potential for clinical application in gastroesophageal reflux disease.