Fat Quantification In Hepatic Steatosis: The Initial Study Of 3D GRE T₂^* Effects Corrected Water-fat Separation Technique In MRI

Objective:To assess the accuracy of a new water-fat separation MR technique which is called volumetric interpolated breath-hold gradient-recalled-echo T₂* effect corrected water-fat separation technique (3D GRE MRI) in fat quantification in live by using 1H-MR spectroscopy (1H-MRS) as the reference standard.Materials and methods:Forty subjects were taken conventional MRI examination of liver which including T1WI, T₂WI, the in-phase and opposed-phase imaging. Two radiologist with MR experience analysed all images and judged whether the subject had hepatic steatosis. Then they were underwent single voxel 1H-MRS and 3D GRE MRI of the liver. The MRI images and MRS spectra features were observed and analyzed. MRS were collected at echo times with 30, 60, 90, 120, and 150 millisecond respectively. The T₂* effects were corrected by using a log-linear least square fitting algorithm. According to the corrected area under the water peak (Swater) and fat peak area (Sfat), the HFFMRS (hepatic fat fraction, HFF) was calculated. The value of HFFMRS was used as the standard to evaluate the accuracy of 3D GRE MRI in diagnosing hepatic steatosis. 3D GRE MRI based on two points DIXON method and used the small flip angle to inhibit T1 effects. In order to correct T₂* effect, one serial in-phase and two serial opposed-phase images were collected. The HFF3D GRE MRI of region of interest (ROI) which was measured directly in hepatic fat fraction gray-scale image. The size and location of ROI were consistant with those of MRS. Serum triglyceride (TG), height and weight were tested at the same time, and body mass index (BMI) was calculated.Results:①The BMI in 40 subjects ranged from 21.20 to 32.49kg/m² with the mean value of 25.46±2.72kg/m². The serum TG ranged from 0.82 to 17.74mmol/L and the median value was 2.39mmol/L. 12 subjects were screened out for normal control group according to BMI <24, serum TG <1.7mmol/L as the normal standard and all of them with no diabetes, hypertension and other risk factors of fatty liver. The rest were divided into the high-risk group.②The HFFMRS in 40 subjects ranged from 0.36% to 17.95% with the median value of 5.05%. The HFFMRS ranged from 0.36% to 3.99% with the median value of 0.59% in control group meanwhile 1.15% to 17.95% with the median value of 8.85% in the high risk group. The HFF3D GRE MRI in 40 subjects ranged from 2.43% to 26.75% with the median value of 10.90%. The HFF3D GRE MRI ranged from 2.43% to 6.19% with the median value of 3.50% in control group meanwhile 3.45% to 26.75% with the median value of 15.38% in the high risk group. The Wilcoxon signed rank test showed the difference between HFF3D GRE MRI and HFFMRS had statistical significance (P<0.01).③The value of HFFMRS and HFF3D RGRE MRI showed a linear correlation. The regression line equation was HFF3D GRE MRI=1.3199 HFFMRS+2.8553(R2=0.9489). The Spearson correlation coefficient was 0.963(P<0.01). The value of HFF3D RGRE MRI was much higher then that of HFFMRS.④When HFF3D RGRE MRI≥11.81%, the new water-fat separation MR technique had diagnostic sensitivity of 1.0, specificity of 0.952, positive predictive value of 89.50%, and negative predictive value of 100% respectively. Conclusion: Initial study showed that 3D GRE T₂* effects corrected water-fat separation technique had high accuracy in hepatic steatosis and had potential on fat-grading in hepatic steatosis.