Application Of Histogram Analysis Of MRI Quantitative Parameters In Pathological Grades Of Gastric Adenocarcinoma

Part 1 Histogram analysis of Dynamic contrast-enhanced MRI for predicting pathological grades in patients with gastric adenocarcinomaObjectiveTo explore the ability of histogram analysis of dynamic contrast-enhanced magnetic resonance imaging（DCE-MRI）in predicting pathological grades of gastric adenocarcinoma.MethodsThe data of 67 patients with gastric adenocarcinoma confirmed by pathological findings after operation was retrospectively collected.All patiens were underwent DCE-MRI examinations before operations.Quantitative parameters of DCE-MRI were obtained after processing,including histogram parameters（mean,10th percentile,25th percentile,50th percentile,75th percentile,90th percentile）of volume transfer constant(K^trans),rate constant(K_ep)and fractional extravascular extracellular space volume（V_e）.These patients were divided into three groups（well,moderately and poorly differentiated）according to the differentiation degree.Using Kruskal-Walls H test to compare the differences among the three groups and Bonferroni method to correct the significance level in pairwise comparison.Receiver operating characteristic curve（ROC）was carried to evaluate the efficacy of the parameters in the diagnosis of poorly differentiated gastric adenocarcinoma.ResultsThere were 32 cases in the poorly differentiated group,30 cases in the moderately differentiated group and 5 cases in the well differentiated group.Among the three groups,the mean,10th percentile,25th percentile,50th percentile,75th percentile,90th percentile of K^transand K_epshowed statistical difference（P<0.05）.The values of mean,10th percentile,25th percentile,50th percentile,75th percentile,90th percentile of K^transand K_epwere significantly higher in poorly differentiated group than those in moderately and well differentiated groups（P<0.05）.The 90th percentile of K^transwas the most effective parameter in diagnosing poorly differentiated gastric adenocarcinoma which the area under the curve（AUC）is 0.853.ConclusionDCE-MRI histogram analysis could be used to assist in evaluating preoperative pathological grades of gastric adenocarcinoma.The 90th percentile of K^transshows the highest efficiency in the diagnosis of poorly differentiated gastric adenocarcinoma.Part 2 The value of histogram analysis of apparent diffusion coefficient in predicting pathological grades of gastric adenocarcinomaObjectiveTo investigate the value of histogram analysis of apparent diffusion coefficient（ADC）in predicting pathological grades of gastric adenocarcinoma.MethodsThe data of 72 patients with gastric adenocarcinoma confirmed by pathology were retrospectively analyzed.Diffusion weighted imaging（DWI）was performed in all patients before surgery.Histogram parameters of the ADC were measured including mean,skewness,kurtosis,1st percentile,10 th percentile,50 th percentile,90 th percentile and 99 th percentile.These patients were divided into well（G1）,moderately（G2）and poorly（G3）differentiated groups according to the postopeative pathology.One-way ANOVA and Kruskal-Walls H test were applied to calculate the differences of the parameters among the three groups.Spearman correlation was used to analyze the relationship between ADC histogram parameters and pathological grades.Using ROC curve to evaluate the diagnostic efficacy of parameters in identifying poorly differentiated gastric adenocarcinoma.ResultsThere were 35 cases in the poorly differentiated group,31 cases in the moderately differentiated group and 6 cases in the well differentiated group of gastric adenocarcinoma.There were significant differences in the mean,1st,10 th,50th,90 th and 99 th percentile of ADC among the three groups（P<0.05）.The mean,1st percentile and 10 th percentile of ADC showed higher diagnostic performance in identifying poorly differentiated gastric adenocarcinoma.The AUCs of mean,1st,10 th percentile of ADC were 0.883,0.869 and 0.902,respectively.The mean and the1 st,10th,50 th,90th and 99 th percentiles of ADC were negatively correlated with pathological grades（the absolute value of r is 0.557 to 0.705,P<0.05）.ConclusionHistogram analysis of ADC can predict the pathological grades of gastric adenocarcinoma and it is hopeful to be an auxiliary and noninvasive way to evaluate the pathological grades.