Repeatability and Reproducibility of Pancreas Volume Measurements Using MRI

Abstract

Reduced pancreas volume, as measured by non-contrast magnetic resonance imaging (MRi), is observed in individuals with newly-diagnosed type 1 diabetes (T1D) and declines over the first year after diagnosis. in this study, we determined the repeatability and inter-reader reproducibility of pancreas volume measurements by MRI. Test-retest scans in individuals with or without T1D (n = 16) had an intraclass correlation coefficient (ICC) of 0.985 (95% CI 0.961 to 0.995) for pancreas volume. Independent pancreas outlines by two board-certified radiologists (n = 30) yielded an ICC of 0.945 (95% CI 0.889 to 0.973). The mean Dice coefficient, a measurement of the degree of overlap between pancreas regions of interest between the two readers, was 0.77. Prandial state did not influence pancreatic measurements, as stomach volume did not correlate with pancreas volume. these data demonstrate that MRi measurements of pancreas volume between two readers are repeatable and reproducible with ICCs that correspond to excellent clinical significance (ICC > 0.9), are not related to changes in stomach volume, and could be a useful tool for clinical investigation of diabetes and other pancreas pathologies.

Deep learning-based pancreas volume assessment in individuals with type 1 diabetes

“…The deep learning-based segmentation calculated smaller pancreas volume in individuals with T1D, in agreement with previous studies using manual segmentation [1][2][3], but absent the subjectivity inherent to manual segmentation. In fact, the agreement between deep learning and manual pancreas segmentation in this study outperformed the agreement between two different readers performed using images derived from the same study [17]. This finding highlights the subjectivity in manual pancreas segmentations which are in turn used to train deep learning models.…”

Development of a standardized MRI protocol for pancreas assessment in humans

“…For example, a meta-analysis of pancreatic volume and fat content found high heterogeneity between studies [ 28 ]. However, we found that MRI of the pancreas performed at a single site is repeatable [ 29 ]. The choice of image acquisition and processing parameters can influence a calculated measure, as previously demonstrated in pancreatic measurements of fat fraction [ 30 ], T1 [ 31 ] and ADC [ 32 ].…”

“…An additional limitation of this study is the use of a single reader to outline all pancreas images in this study. The use of multiple readers may lead to variation in MRI measurements of the pancreas [ 29 ]. Magnetic resonance elastography is another promising technique for assessing the pancreas, but as it requires specialized hardware and software not available at our study sites it was not included in this study.…”

Quantitative Magnetic Resonance Imaging of the Pancreas of Individuals With Diabetes

“…Pancreas borders can be difficult to delineate, which may lead to reader subjectivity in pancreas volume measurements. However, reproducibility across multiple readers reveals good agreement between two readers outlining the same image (24,25). One caveat is that pancreas volume changes over the course of the human lifespan in the absence of disease, with rapid increases in size over childhood and pancreas atrophy in later life (14).…”

“…Porcine pancreas volume calculated from MRI demonstrated excellent correlation with gold standard water displacement measurements (23), indicating the accuracy of the technique. Furthermore, in human subjects who received MRI scans in quick succession, measurements of pancreas volume were highly repeatable (24,25). Pancreas borders can be difficult to delineate, which may lead to reader subjectivity in pancreas volume measurements.…”