The Research Of Different Dynamic Enhanced MRI Models In The Differential Diagnosis From HCC To Liver Metastases

Objective:This study mainly analyzed the value of histogram parameters of dynamic enhanced magnetic resonance(DCE MRI)Exchange model and Extended Tofts model in the differential diagnosis of hepatocellular carcinoma(HCC)and rectal carcinoma with liver metastases.Methods and materials:After screening for inclusion criteria and exclusion criteria,MRI scan data were collected from 77 patients with HCC and rectal carcinoma with liver metastases from July 2016 to December 2017,including 55 cases of HCC and 22 cases of rectal carcinoma with liver metastases.Preoperative DCE-MRI examination was performed in 77 patients,the obtained data use GE medical image post-processing software Omni Kinetics to analysis and processing,respectively in Exchange model and Extended Tofts model two different pharmacokinetic model to calculate the corresponding quantitative parameters:the hepatic perfusion index(HPI),volume transfer constant(Ktrans),extracellular back to plasma speed constant(Kep),extravascular extracellular space volume fraction(Ve)and plasma volume fraction(Vp).Then,the histogram was used to analyze each quantitative parameter and record the corresponding histogram parameters.The mean value,10%digit,25%digit,50%digit,75%digit and 90%digit were selected for statistical analysis using the statistical software of SPSS version 19.When the data results meet the normal distribution and reach the homogeneity of variance,the form of "mean± standard deviation" is used to analyze the difference between the two groups of values by independent sample t-test.Data that cannot meet the normal distribution or reach the homogeneity of variance are represented by the form of "median(upper and lower quartile)" and analyzed by mann-whitney U test.P<0.05 means the data difference is statistically significant.Finally,ROC curve was used to analyze the diagnostic efficacy of the data with statistical differences in the histogram parameters of the two models.Results:In the Extended Tofts model,there was a statistical difference between the HCC group and the rectal cancer group in the histogram parameters Ktrans mean,ktrans50%number,ktrans75%number,ktrans90%number,Kep mean,kep75%number,kep90%number,ve10%number,Vp mean,vp50%number,vp75%number,vp90%number,HPI mean,hpi10%number,hpi25%number,hpi75%number,hpi75%number,hpi75%number and hpi90%number.The mean values of Ktrans,Kep,Vp,HPI and high percentiles of 75%and 90%digits of HCC tumor tissues were significantly higher than those of colorectal cancer,which could be used for differential diagnosisIn the Exchange model,the histogram quantitative parameters of the HCC group and the colorectal cancer group included Ktrans mean,ktrans10%digit,ktrans25%digit,ktrans50%digit,ktrans75%digit,ktrans75%digit,ktrans90%digit,Ve mean,ve10%digit,ve25%digit,Vp mean,vp10%digit,vp50%digit,HPI mean,hpi10%digit,hpi25%digit,hpi75%digit,hpi75%digit and hpi90%digitStatistical difference.The mean values of Ktrans,Ve,Vp,HPI and the low percentiles of 10%and 25%digits of HCC tumor tissues were significantly higher than those of colorectal cancer,which could be used for differential diagnosisThe ROC curves of histogram quantitative parameters with statistical differences between the two models were drawn,and the HPI mean AUC in the Extended Tofts model was the largest(0.963),the sensitivity was 92.59%,and the specificity was 92.00%.In the Exchange model,the AUC of ktrans90%digit is the largest,which is 0.936,the sensitivity is 94.44%,and the specificity is 88.00%Conclusion:The mean values of Ktrans,Kep,Vp,HPI and high percents in the Extended Tofts model can be used for the differential diagnosis of liver metastases from HCC to rectal cancer.The mean values of Ktrans,Ve,Vp,HPI and low hundreds in the Exchange model can be used for the differential diagnosis of HCC and colorectal carcinoma with liver metastases.The two models Ktrans and HPI had the best diagnostic value,among which the HPI mean value of Extended Tofts model and ktrans90%digit value of Exchange model were the most effective in differential diagnosis.