Histogram Analysis of Diffusion Kurtosis Magnetic Resonance Imaging for Diagnosis of Hepatic Fibrosis

Abstract

ObjectiveTo investigate the diagnostic value of diffusion kurtosis imaging (DKI) histogram analysis in hepatic fibrosis staging.Materials and MethodsThirty-six rats were divided into carbon tetrachloride-induced fibrosis groups (6 rats per group for 2, 4, 6, and 8 weeks) and a control group (n = 12). MRI was performed using a 3T scanner. Histograms of DKI were obtained for corrected apparent diffusion (D), kurtosis (K) and apparent diffusion coefficient (ADC). Mean, median, skewness, kurtosis and 25th and 75th percentiles were generated and compared according to the fibrosis stage and inflammatory activity.ResultsA total of 35 rats were included, and 12, 5, 5, 6, and 7 rats were diagnosed as F0–F4. The mean, median, 25th and 75th percentiles, kurtosis of D map, median, 25th percentile, skewness of K map, and 75th percentile of ADC map demonstrated significant correlation with fibrosis stage (r = −0.767 to 0.339, p < 0.001 to p = 0.039). The fibrosis score was the independent variable associated with histogram parameters compared with inflammatory activity grade (p < 0.001 to p = 0.041), except the median of K map (p = 0.185). Areas under the receiver operating characteristic curve of D were larger than K and ADC maps in fibrosis staging, although no significant differences existed in pairwise comparisons (p = 0.0512 to p = 0.847).ConclusionCorrected apparent diffusion of DKI histogram analysis provides added value and better diagnostic performance to detect various liver fibrosis stages compared with ADC.

Differentiating mild and substantial hepatic fibrosis from healthy controls: a comparison of diffusion kurtosis imaging and conventional diffusion-weighted imaging

“…In theory, the MK reflects the tissue complexity (36–38). In our study, MK values showed no significant differences between healthy controls and those with S1 or S2 fibrosis, and had no correlation with the fibrosis stages, which was inconsistent with previous clinical and animal studies (24–29). However, the MK value in the same fibrosis stage was inconsistent across different studies, and the different values of MK in different fibrosis stages are very small.…”

“…Our results showed that the MD values decreased as the fibrosis advanced, which is consistent with the pathological changes of liver fibrosis, a complex assembly of collagen fibers, glycosaminoglycans, and proteoglycans in liver parenchyma that restricted the molecular diffusion, and the results of previous animal and clinical studies (24–29). MD values could differentiate S0 from S1 and S0 from S2, had an excellent diagnostic accuracy for predicting S1–2, and a moderate diagnostic accuracy for predicting S2.…”

“…Diffusion kurtosis imaging (DKI) is an advanced DWI model that quantifies the non-Gaussian behavior of diffusion by using high b values and kurtosis analysis. Previous studies of DKI in animal models and clinical patients have reported the possibility of staging liver fibrosis, but results regarding whether DKI provided similar or better diagnostic performance compared to cDWI were inconsistent (24–29). In addition, the diagnostic performance of mean apparent diffusion (MD) and mean kurtosis (MK) derived from DKI for liver fibrosis was different, especially MK (24–29).…”

Histogram analysis in prostate cancer: a comparison of diffusion kurtosis imaging model versus monoexponential model

“…Usually only mean values are obtained by routine measurements, but mean values cannot reflect the underlying distribution of signal intensity of the whole tumor (13). Prostate cancer is heterogeneous; tumors can consist of cancerous tissue interspersed with normal tissue.…”

“…Therefore, the heterogeneity of prostate cancer would cause artificially high mean ADC values (14). Histogram analysis is a mathematical method that may provide the distribution of signal intensity of voxels, and this method can help us more comprehensively estimate intratumor heterogeneity (13). In recent years, histogram analysis of a monoexponential DWI model has been conducted in prostate cancer.…”