Meeting Banner
Abstract #3427

Automated Liver Stiffness Measurements with Magnetic Resonance Elastography

Bogdan Dzyubak1, Kevin Glaser2, Meng Yin2, Armando Manduca2, Richard Ehman2

1Mayo Graduate School, Mayo Clinic, Rochester, MN, United States; 2Radiology, Mayo Clinic, Rochester, MN, United States

Measurements of liver stiffness based on MR Elastography images have an inherent variability due to differences in ROI definition between readers. This study was aimed at developing an automatic algorithm for segmenting the liver in MRE images and measuring stiffness in an ROI with reliable wave propagation. The algorithm was shown to perform as well as an experienced MRE reader even in images with high artifact. This technique may streamline the measurement of liver stiffness while reducing time, variability, and cost.

Keywords

adaptive additionally analyzed approximate arbitrary around artifact artifacts assessment attempted automated automatically basic bland blood blue blurring caused changing cleaned cleaning clinic cohort cold collected common comparable comparing comprised confidence consistent contained containing contour correlation created currently default desktop detect detectable detection differed diffuse done drawing dropout easily edges editors element eroded erratic every exam excluded exclusion expands experience experienced exterior fast fibrosis find force frequency gradient green hepatic highly histogram hole homogeneous increasingly inherent initial initialization initialized intensity intraclass inversion leakage least leave liver location made magnitude majority manual manually many mask mayo measures morphological morphologically normalized objects oblique organs original outputs patient peak peaks pixel polynomial practice preliminary processing propagation proposed quality reader readers reasonable relatively reliable removal removes removing require review school segmentation separate shape slice slices spin spots stages stiffness stiffnesses structures subset though threshold thresholded took trained training tumors variance vessel vessels wave years yield yielded