Predicting Histological Grade of HCC in Rats using Intravoxel Incoherent Motion Imaging

ObjectiveTo predict histological grade of hepatocellular carcinoma (HCC) in Wistar rats using intravoxel incoherent motion imaging (IVIM) with magnetic resonance imaging (MRI). MethodsLiver lesions of HCC rat models induced by oral diethylnitrosamine (DEN) were scanned by IVIM to obtain apparent diffusion coefficient (ADC) and IVIM parameters, including true diffusion coefficient (D), pseudodiffusion coefficient (D*) and perfusion fraction (f). These HCC lesions were confirmed by Hematoxylin-Eosin (HE) stain and pathologically graded into low (grade Ⅰ+Ⅱ) and high (grade Ⅲ+Ⅳ) using the Edmondson-Steiner method. The ADC and IVIM parameters of the two grade groups were compared: their diagnostic performance were assessed using ROC curves. ResultsHCC models were successfully established in 48 rats, containing 50 HCC lesions (28 low-grade and 22 high-grade). The high-grade lesions had lower ADC (P=0.009) and D (P=0.005) values and higher D* (P=0.032) and f (P=0.044) values compared with the low-grade lesions. The largest jonden index appeared in the ROC curves at 0.907 8×10^-3 mm²/s of ADC, 0.817 6×10^-3 mm²/s of D, 24.31×10^-3 mm²/s of D*, and 14.4% of f, respectively. The area under curves (AUCs) of these parameters ranged from 0.5 to 0.9, showing no significant differences (P>0.05). ConclusionADC and IVIM parameters have equal and moderate diagnostic values in predicting histologic grade of HCCs，which can be used for estimating pathological grading of HCCs before surgery.