Are the thickness and ADC values reliable to evaluate pancreas in children? A retrospective MRI study.

Abstract

To define the pancreatic thickness on the transverse plane and apparent diffusion coefficient (ADC) values at multiple anatomical locations in the normal pancreas of children and to evaluate inter-rater and intra-rater reliability. Two reviewers measured the thickness and ADC values of the pancreas at four locations obtained from two different 1,5 Tesla MRI scanners [MAGNETOM Symphony (Siemens Medical Systems, Erlangen, Germany) and Signa HDxt(GE Healthcare, Milwaukee, WI, USA)]. Measurements were made based on 190 retrospective MRI examinations. Patients were distributed among ten age groups for thickness measurements and three age groups for ADC measurements between 0 and 18years. The thickness measurements from each segment of the pancreas were significantly correlated with patient age, height and weight (p < 0.01). Inter-rater agreement was excellent for pancreatic head (intraclass correlation coefficient (ICC): 0.837) and good for pancreatic neck, body and tail (ICC: 0.646, 0.632 and 0.678, respectively). We also defined the mean ADC values for three age groups. There was significant difference in the mean ADC value of the head and body by both the vendor and age group (p < 0.05). The mean ADC values and mean ranks of the Signa HDxt were significantly higher and the frequencies were lower than those of the MAGNETOM Symphony at all locations (p < 0.001). The thickness of the pancreatic head is more reliable than that at other locations. Pancreatic body and tail measurements are not as reliable as head or neck measurements due to discrepancies between published studies and low interobserver agreement. ADC values of the normal pancreas can be significantly different among vendors; therefore, ADC comparison on follow-up should be performed using same MRI machine.