Value of diffusion weighted imaging based on monoexponential and biexponential model in the differential diagnosis between benign and malignant liver neoplasms / 中华放射学杂志

ABSTRACT

Objective To investigate the utility value of monoexponential and biexponential DWI in the differential diagnosis between benign and malignant liver neoplasms.Methods Seventy three patients with pathologically or clinically confirmed liver mass,were analyzed retrospectively and categorized into benign and malignant groups between January 2013 and October 2013.Malignant group included 46 patients with 53 lesions,while 27 patients in benign group had 35 lesions.All patients underwent MR examinations on 3.0T system (GE 750).Conventional MR T1WI,T2WI,DWI(b=0,800 s/mm2) (to obtain ADC with monoexponential modeling),multi-b value DWI(b=0,20 50,100,200,400,600,800 and 1 200 s/mm2) (to obtain Slow-ADC,Fast-ADC,f with biexponential modeling) and dynamic enhancement were performed.The ADC,Slow-ADC,Fast-ADC and f mean values of benign and malignant liver neoplasms were measured and analyzed by using independent samples t test.Diagnostic efficacy of these parameters in malignant group was evaluated by using receiver operating characteristic curve,with histopathologic findings as the gold standard.Results ADC,Slow-ADC,Fast-ADC and f of malignant group were lower than those of benign group [ADC(1.79±0.35)× 10-3 mm2/s vs (1.16±0.36) × 10-3 mm2/s; Slow-ADC(1.67±0.25) × 10-3 mm2/s vs(0.94±0.22)×10-3mm2/s; Fast-ADC(72.40±23.70)×10-3mm2/s vs(34.62±17.43)×10-3mm2/s; and f(33.59± 11.77)％ vs (22.28±8.97)％ in benign and malignant groups,respectively).Significant inter-group difference was observed in ADC,Fast-ADC,Slow-ADC and f (t=0.89,14.77,8.96 and 5.47,respectively and P＜0.05).The areas under the ROC curve (AUC) of ADC,Slow-ADC,Fast-ADC and fwere 0.938,0.974,0.895 and 0.789,respectively.The sensitivity and specificity of ADC,Slow-ADC,Fast-ADC and fwere 90.6％ (48/53),96.2％ (51/53),90.6％ (48/53) and 90.6％ (48/53) and 85.7％ (30/35),91.4％ (32/35),82.9％ (29/35) and 57.1％ (20/35)respectively for differentiating benign from malignant hepatic lesions.Conclusion ADC obtained with mono-exponential modeling and Fast-ADC,Slow-ADC,f obtained with biexponential modeling are useful parameters in distinguishing benign and malignant hepatic lesions,among which slow-ADC demonstrates the highest diagnostic efficacy.