Reproducibility and Repeatability of Intravoxel Incoherent Motion MRI Acquisition Methods in Liver

• Published in: Journal of magnetic resonance imaging Vol. 60; no. 4; pp. 1691 - 1703
• Main Authors: Vasquez, Juan A., Brown, Marissa, Woolsey, Mary, Abdul‐Ghani, Mohammad, Katabathina, Venkata, Deng, Shengwen, Blangero, John, Clarke, Geoffrey D.
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 01.10.2024; Wiley Subscription Services, Inc
• Subjects: Correlation coefficient; Correlation coefficients; Data acquisition; diffusion; Diffusion coefficient; Diffusion rate; Distortion; Family studies; Fatty liver; Fibrosis; Field strength; Image acquisition; Image quality; intervoxel incoherent motion; Liver; Magnetic resonance imaging; Parameter estimation; Reproducibility; Statistical analysis; Statistical tests; Steatosis; Variability; diffusion; fibrosis; liver; intervoxel incoherent motion
• ISSN: 1053-1807; 1522-2586; 1522-2586
• DOI: 10.1002/jmri.29249

Background Intravoxel incoherent motion (IVIM) diffusion weighted MRI (DWI) has potential for evaluating hepatic fibrosis but image acquisition technique influence on diffusion parameter estimation bears investigation. Purpose To minimize variability and maximize repeatably in abdominal DWI in terms of IVIM parameter estimates. Study Type Prospective test–retest and image quality comparison. Subjects Healthy volunteers (3F/7M, 29.9 ± 12.9 years) and Family Study subjects (18F/12M, 51.7 ± 16.7 years), without and with liver steatosis. Field Strength/Sequence Abdominal single‐shot echo‐planar imaging (EPI) and simultaneous multi‐slice (SMS) DWI sequences with respiratory triggering (RT), breath‐holding (BH), and navigator echo (NE) at 3 Tesla. Assessment SMS‐BH, EPI‐NE, and SMS‐RT data from twice‐scanned healthy volunteers were analyzed using 6 × b‐values (0–800 s⋅mm−2) and lower (LO) and higher (HI) b‐value ranges. Family Study subjects were scanned using SMS and standard EPI sequences. The biexponential IVIM model was used to estimate fast‐diffusion coefficient (Df), fraction of fast diffusion (f), and slow‐diffusion coefficient (Ds). Scan time, estimated signal‐to‐noise ratio (eSNR), eSNR per acquisition, and distortion ratio were compared. Statistical Tests Coefficients of variation (CoV) and Bland Altman analyses were performed for test–retest repeatability. Interclass correlation coefficient (ICC) assessed interobserver agreement with P < 0.05 deemed significant. Results Within‐subject CoVs among volunteers (N = 10) for f and Ds were lowest in EPI‐NE‐LO (11.6%) and SMS‐RT‐HI (11.1%). Inter‐observer ICCs for f and Ds were highest for EPI‐NE‐LO (0.63) and SMS‐RT‐LO (0.76). Df could not be estimated for most subjects. Estimated eSNR (EPI = 21.9, SMS = 4.7) and eSNR time (EPI = 6.7, SMS = 16.6) were greater for SMS, with less distortion in the liver region (DR‐PE: EPI = 23.6, SMS = 13.1). Data Conclusion Simultaneous multislice acquisitions had significantly less variability and higher ICCs of Ds, higher eSNR, less distortion, and reduced scan time compared to EPI. Evidence Level 2 Technical Efficacy Stage 1