• Takuya Hinoda, Yasutaka Fushimi, Tomohisa Okada, Koji Fujimoto, Chunlei Liu, Akira Yamamoto, Tsutomu Okada, Aki Kido, and Kaori Togashi.
• From the *Department of Diagnostic Imaging and Nuclear Medicine, Kyoto University Graduate School of Medicine, Kyoto, Japan; and †Brain Imaging and Analysis Center and Department of Radiology, Duke University Medical Center, Durham, NC.
• Invest Radiol. 2015 Aug 1; 50 (8): 522-30.

ObjectivesThe aim of this study was to assess the consistency and reproducibility of quantitative susceptibility mapping (QSM) at 3-T and 1.5-T magnetic resonance (MR) scanners.Materials And MethodsThis study was approved by institutional ethics committee, and written informed consent was obtained. Twenty-two healthy volunteers underwent 2 examinations on different days. Each examination consisted of MR imaging on both 3-T and 1.5-T MR scanners. The data from both scanners and examination days were obtained, and QSM was calculated with STI Suite using 2 different algorithms--harmonic phase removal using laplacian operator (HARPERELLA) and a sophisticated harmonic artifact reduction for phase data (SHARP) method with a variable radius of the spherical kernel at the brain boundary (V-SHARP). We evaluated consistency of QSM between 3 T and 1.5 T and the reproducibility between the first and second examinations using 2-phase processing methods (HARPERELLA and V-SHARP).ResultsSusceptibility values of regions of interests at 3 T were highly correlated with those at 1.5 T with good agreement (HARPERELLA, R2 = 0.838; V-SHARP, R2 = 0.898) (average difference, ±1.96 SD; HARPERELLA, -0.012 ± 0.046; V-SHARP, -0.002 ± 0.034). Reproducibility analysis demonstrated excellent correlation between the first and second examination at both 3 T and 1.5 T for both algorithms (HARPERELLA at 3 T, R2 = 0.921; 1.5 T, R2 = 0.891; V-SHARP at 3 T, R2 = 0.937; 1.5 T, R2 = 0.926). Bland-Altman analysis showed excellent reproducibility for HARPERELLA (3 T, -0.003 ± 0.032; 1.5 T, -0.003 ± 0.038) and V-SHARP (3 T, -0.003 ± 0.027; 1.5 T, -0.003 ± 0.029). Susceptibility values of these 2 algorithms were highly correlated with good agreement (3T, R2 = 0.961; 1.5 T, R = 0.931) (3 T, 0.009 ± 0.023; 1.5 T, -0.003 ± 0.049).ConclusionsQuantitative susceptibility mapping with HARPERELLA and V-SHARP demonstrated good reproducibility at 3 T and 1.5 T, and QSM with V-SHARP demonstrated good consistency at 3 T and 1.5 T.

20 keywords...

hide…