Quantitative characterization of carotid plaque components using MR apparent diffusion coefficients and longitudinal relaxation rates at 3T: A comparison with histology

• Published in: Journal of magnetic resonance imaging Vol. 48; no. 6; pp. 1657 - 1667
• Main Authors: Ota, Hideki, Tamura, Hajime, Itabashi, Ryo, Yazawa, Yukako, Nakamura, Yasuhiro, Hisamatsu, Kenji, Takamatsu, Manabu, Endo, Hidenori, Niizuma, Kuniyasu, Enomoto, Yukiko, Nagasaka, Tatsuo, Kajita, Kimihiro, Watanabe, Mika, Yoshimura, Shinichi, Yuan, Chun
• Format: Journal Article
• Language: English
• Published: United States Wiley Subscription Services, Inc 01.12.2018
• Subjects: Alcohols; apparent diffusion coefficient; carotid plaque components; Classification; Diffusion rate; Field strength; Gadolinium; Hemorrhage; Histology; Ischemia; Lipids; longitudinal relaxation rate; Magnetic resonance imaging; Mathematical analysis; Medical imaging; Plaques; Polyvinyl alcohol; Probability density functions; Statistical analysis; Statistical tests; apparent diffusion coefficient; magnetic resonance imaging; histology; carotid plaque components; longitudinal relaxation rate
• ISSN: 1053-1807; 1522-2586; 1522-2586
• DOI: 10.1002/jmri.26216

Background There is limited evidence of parametric MR mapping to characterize carotid plaques associated with cerebral ischemic events. Purpose To explore the apparent diffusion coefficients (ADCs) and longitudinal relaxation rates (R1) of carotid plaques, including areas of hemorrhage, lipid‐rich/necrotic core (LR/NC) without hemorrhage, and fibrous tissue (Fbr) Study Type Prospective. Subjects Twelve patients who underwent carotid endarterectomy. Field Strength/Sequence R1 was measured using double angle Look–Locker acquisition on 3T systems. Single‐shot spin‐echo echo‐planar imaging with fat suppression and outer‐volume suppression (OVS‐DWEPI) with b values of 10 and 500 s/mm2 was used for diffusion‐weighted imaging. Assessment A phantom study using diluted gadolinium solutions and polyvinyl alcohol solutions was used to validate the two protocols. Regions of interest (ROIs) were manually outlined on MR images for areas of LR/NC, hemorrhage, and Fbr based on histological cross‐sections. Pixel‐based R1 and ADC values in the ROIs were plotted for each component. The probability density function of the plots determined the optimum contours to separate the three components in the ADC‐R1 plane. The LR/NC, hemorrhage, and Fbr regions were mapped on MR images based on the above results and compared to histological results. Statistical Tests The R1 values of the phantom measurements were tested using Bland–Altman analysis. The accuracies of the MRI classification were calculated. Results R1 values <8 s−1 calculated using our method agreed with those calculated using an inversion‐recovery fast‐spin‐echo sequence (error, ≤0.1 s−1). ADC values obtained using OVS‐DWEPI were 4.1% higher than those obtained using standard echo‐planar imaging. LR/NC (R1, 0.4–1.2 s−1; ADC, 0–1.5 μm2/ms), hemorrhage (R1 ≥ 1.5 s−1; ADC, 0.5–1.5 μm2/ms), and Fbr (R1, 0.2–0.8 s−1; ADC, 1.5–2.9 μm2/ms) were separated on the plots. The accuracies of MRI classification were LR/NC, 0.86; hemorrhage, 0.79; and Fbr, 0.77. Conclusion The combination of ADC and R1 values measured using our method enabled differentiation among LR/NC, hemorrhage, and Fbr. Level of Evidence: 1 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2018;48:1657–1667