Introduction

Liver fibrosis is a hallmark of chronic liver disease (CLD) characterized by the excessive accumulation of extracellular matrix proteins (for fibrogenesis) [1]. Liver fibrosis may progress to cirrhosis, the end stage, which constitutes the most important risk factor for developing hepatocellular carcinoma [2]. Thus, early detection and treatment of the underlying cause of liver disease is critical because liver transplantation constitutes the only curative therapy for decompensated liver cirrhosis.
Although liver biopsy is the gold standard for diagnosis and staging of liver fibrosis, the method has some limitations, including sampling errors, low patient acceptance and complications such as pain, bleeding, infection and rarely death [3]. Hence, there have been considerable efforts to develop noninvasive imaging techniques and quantification programs for diagnosis, staging, and monitoring of liver fibrosis. Recently, several quantification softwares in liver diseases have been introduced for assessing the clinical indication of liver fibrosis using MR images. Heterogeneity program using coefficient of variation (CV) map can help to grade the fibrosis severity in the patients with chronic hepatitis B [4]. Also, liver surface nodularity program using multi-polynomial curve fitting can be useful to differentiate the fibrosis severity in the nonalcoholic fatty liver disease (NAFLD) [5]. Therefore, liver heterogeneity and surface nodularity scores can serve important information to differentially diagnose liver fibrosis in the liver diseases.
In this study, we developed a MRI-suitable quantification software for assessing liver heterogeneity and nodularity and compared hepatic heterogeneity and nodularity according to fibrosis grades in CLD.

Subjects and Methods

Liver heterogeneity (L_Het) and (L_Nod) nodularity quantification program was coded by Matlab (ver. 2016a). The processing procedures (L_Het and L_Nod) were as follows: bias correction, semi-automatic boundary detection, liver segmentation and L_Het & L_Nod measurements in the segmented liver (Fig. 1).
The study protocol was approved by the institutional review board (IRB) of University Hospital. A total of 90 patients with CLD divided to three groups according to the serum biomarkers of fibrosis-4 index (FIB-4, equation I) values as follows: G1, non-significant fibrosis group <1.45; G2, significant fibrosis group 1.45-3.25; and G3, advanced fibrosis group >3.25 (see Table 1).
FIB-4 = (Age [year] × AST [U/L] ) / (platelet count [10⁹/L] × square root of ALT[U/L]) -- Eq. I
* The upper limit of normal (ULN) AST was 35 in this study.
All MRI examinations were performed on a 3T MRI scanner (Achieva; Philips, Netherlands) with a 32-channel receiver body matrix coil. The T1WI (repetition time [TR]/TE= 4.2/1.97 ms) were acquired with three-dimensional T1 high-resolution isotropic volume excitation pulse sequence: field of view (FOV)= 38×38×14 cm³, matrix size= 512×512, number of excitation (NEX)= 2, slice thickness= 0.74×0.74×2.0 mm³, number of slices= 70 and scan time= 14 sec. Together with a single breath-hold technique, the imaging plane in the upper abdomen was axial and the sequence was triggered to expiration. Additional multislice T1- and T2-weighted sequences were obtained in sagittal, axial and coronal planes as the liver MRI protocol. Total examination time ranged from 35–45 minutes.
Our L_Het and L_Nod quantification program was used the MR images of DICOM format to generate the heterogeneity and nodularity scores using previously described procedure [4, 5]. The MR images were analyzed from the two radiologists and L_Nod scores were generated to assure that no sharp turns were falsely provided by the L_Nod program. At least three and/or four ROI measurements were performed for each subject. A final L_Nod score was calculated by the program as an averaged value of the individual measurements, with a higher L_Nod score indicating a higher degree of surface nodularity. L_Het and L_Nod scores among fibrosis grades (G1-G3) were compared by using ANOVA with Tukey’s post-hoc test.

Results and Discussion

The liver heterogeneity and nodularity quantification was extracted as a reference line and two radiologists (with more than 10 years of experience) finally confirmed the liver surface line (Fig. 2). Mean L_Het scores in three groups were G1 5.52±1.01, G2 6.78±1.24 and G3 7.55±2.16 (p<0.001; Table 2). In multiple comparison, L_Het scores are significantly different from each other (G1 vs G2 (p=0.001); G1 vs G3 (p<0.001); and G2 vs G3 (p=0.001)). Mean L_Nod scores in three groups were G1 0.95±0.09, G2 1.12±0.12 and G3 1.15±0.17 (p<0.001; Table 2). In multiple comparison, L_Nod scores are significantly different in G1 vs G2 (p<0.001) and G1 vs G3 (p<0.001), whereas not significant difference between G2 vs G3 (p=0.287). Thus, the L_Het and L_Nod quantification can be a non-invasive technique capable of detecting fibrotic changes in the liver of CLD. Especially, compared with L_Nod score, L_Het score quantified from our CLD data focusing on fibrosis grades of CLD is a useful quantitative imaging biomarker that can be used to diagnose and stage hepatic fibrosis.

Conclusion

This study developed a MRI-suitable liver heterogeneity and nodularity quantification software. The findings demonstrate that the quantitative heterogeneity and nodularity scores can help to differentially diagnose fibrosis stage in CLD using clinical MR images.

Acknowledgements

This study was supported by the grants of the National Research Foundation of Korea (NRF) (2016M3A9A7918501) and the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare (HI18C1216).