Inter‐observer reproducibility and analysis of gastric volume measurements and gastric emptying assessed with magnetic resonance imaging

Abstract

Magnetic resonance (MR) imaging provides direct, non-invasive measurements of gastric function and emptying. The inter-observer variability (IOV) of MR volume measurements and the most appropriate analysis of MR data have not been established. To assess IOV of total gastric volume (TGV) and gastric content volume (GCV) measurements from MR images and the ability of standard power exponential (PowExp), and a novel linear exponential (LinExp) model to describe MR data. Ten healthy volunteers received three different volumes of a liquid nutrient test meal (200-800 mL) on 3 days in a randomized order. Magnetic resonance scans were acquired using a 1.5T system every 1-5 min for 60 min. Total gastric volume and GCV were measured independently by three observers. Volume data were fitted by PowExp and LinExp models to assess postprandial volume change and gastric emptying half time (T(50) ). An initial rise in GCV and TGV was often observed after meal ingestion, thereafter GCV and TGV decreased in an approximately linear fashion. Inter-observer variability decreased with greater volumes from 12% at 200 mL to 6% at 600 and 800 mL. Inter-observer variability for T(50) was <5%. PowExp and LinExp models provided comparable estimates of T(50) ; however, only LinExp described dynamic volume change in the early postprandial period. Gastric MR provides quantitative measurements of postprandial volume change with low IOV, unless the stomach is nearly empty. The novel LinExp model describes the dynamic volume changes in the early postprandial period more accurately than the PowExp model used in existing gastric emptying studies.