T1 signal intensity ratio variability based on sampling strategies in the pancreas of children and young adults.

Abstract

Purpose: T1-weighted signal intensity ratios (SIR) comparing pancreas to spleen (SIRps) or muscle (SIRpm) can semiquantitatively assess T1 signal change associated with pancreatitis. However, there is no standardized methodology for generating these ratios. We set out to determine the impact of MRI sequence as well as region of interest (ROI) location, shape, and size on T1 SIR.

Methods: Retrospective analysis of T1-weighted MR images from 118 patients acquired 2018-2023. A single observer placed ovoid ROIs in the pancreas body/tail and head/uncinate, spleen, and left erector spinae muscle and large irregular ROIs in the pancreas tail and spleen. ROIs were placed on images from two sequences: 3D radial 2 point mDIXON RF spoiled gradient recalled echo sequence (radial) and breath-hold 3D 2-point mDIXON RF spoiled gradient echo (BH). T1 SIR were calculated from mean signal intensity, and agreement was calculated with intraclass correlations coefficients (ICC) and Bland-Altman difference analyses.

Results: 118 participants, 57 (48%) female, with mean age 13.7 ± 5.6 years (48%) were included. Agreement was good for SIRps based on irregular versus round ROIs (radial: ICC = 0.90; BH: ICC = 0.91). Agreement was moderate for SIR based on sampling the pancreas body/tail versus head/uncinate (ICC = 0.67-0.76) and poor to moderate based on reference organ (muscle vs. spleen) (ICC = 0.41-0.61). Between sequences, agreement was moderate (ICC = 0.55-0.72, mean difference 0.04-0.09).

Conclusion: The size and shape of the ROI used to sample the pancreas does not meaningfully change T1 SIR but the location sampled, the reference organ used, and the MRI sequence used meaningfully change T1 SIR, potentially impacting disease diagnosis and staging.