Purpose or Case Report: To compare liver shear stiffness estimates obtained by Automatic Liver Elasticity Calculation (ALEC) processing of two-dimensional (2D) gradient-recalled echo (GRE) magnetic resonance elastography (MRE) data to stiffness values obtained from standard-of-care manual processing.
Methods & Materials: 62 patients with autoimmune liver diseases (sclerosing cholangitis or autoimmune hepatitis) underwent 2D GRE MRE at 60 Hz on a 1.5T scanner (Ingenia; Philips Healthcare) for this IRB-approved study. For manual processing, elastograms were generated in Intellispace Portal v10.1 (Philips Healthcare) with two readers (R1, R2) independently placing regions-of-interest (ROIs) to calculate mean liver stiffness (weighted for ROI size). ALEC (Mayo Clinic; Rochester, MN) generated elastograms with automatically segmented ROIs directly from MRE magnitude and phase images. Intellispace portal and ALEC both used a Multimodal Direct Inversion (MMDI) algorithm for computation of stiffness maps. Results were compared using Pearson’s correlation (r), intra-class correlation coefficients (ICC), and Bland-Altman analyses.
Results: Study participants were in the age range of 7-20 years (mean age: 15.5 years; 33 males). ICC for inter-reader (R1 vs. R2) agreement for manual processing was excellent (ICC=0.989, 95%CI: 0.981-0.993) with mean bias of -0.05 kPa (95% Limits of Agreement (LoA): 0.47 to -0.57 kPa). Correlation between manual and ALEC processing was excellent (R1: r=0.987; R2: r=0.983) with mean bias of -0.33 kPa (R1 95% LoA: 0.36 to -1 kPa) and -0.28 kPa (R2 95% LoA: 0.42 to -0.97 kPa). ALEC measurements were perfectly reproducible (r=1; mean bias=0 kPa, 95% LoA: 0 kPa). Manual processing took 5:48 ± 1:07 minutes per subject. All the data was processed automatically with ALEC in the background using batch file that took 3:30 minutes to set up. Mean computation time for each case was 2:53 mins on a Microsoft Windows 10 system with Intel® Core ™ i5-7500, 4 Core CPU and 16 GB RAM.
Conclusions: Liver stiffness values computed using ALEC showed excellent agreement with manual analysis and were perfectly reproducible. ALEC processing of MRE should eliminate inter-observer variability, facilitate workflows, and reduce post processing time.