The Value Of DKI Derived Parameters In The Diagnosis Of Rectal Adenocarcinoma At 3.0T MR

Part one.Diffusion kurtosis imaging study of rectal adenocarcinoma at 3.0 Tesla magnetic resonance :effect of region of interest on measurement consistency of parametersPurpose: To explore the effect of region of interest(ROI)on measurement consistency of parameters of diffusion kurtosis imaging study of rectal cancer at 3.0 Tesla magnetic resonance.Materials and Methods: Our study is a retrospective study.Fifty patients with rectal cancer were involved in this study.Diffusion kurtosis imaging was performed before surgery.Kurtosis and diffusivity from DKI in patients with rectal cancer were measured by two radiologists with three ROI protocols: small solid samples 、 single-slice and whole-volume.Two radiologists measured the parameters,and one radiologist measured the parameters again after one month.IBM SPSS Statistics 20 and Med Calc v13.1.2.0 were used to analysis the data.Parameters(kurtosis and diffusivity)from three protocols were tested by using the Paired Sample T test.The three protocols were compared for intra-and inter-observer variability by calculating the interclass correlation coefficient(ICC)and Bland-Altman plots.Results: Diffusivity from three ROI protocols were not statistically significant different(t=0.041、0.673、1.511,P=0.967、0.504、0.137).Kurtosis from the single slice method and whole-volume method were not statistically significant different(t=1.532,P=0.132).Kurtosis measured by the small solid samples method and single slice method,and kurtosis measured by the small solid samples method and whole-volume method were statistically significant different(t=3.631、4.229,P=0.001、<0.001).The ICC values of diffusivity from two radiologists measured by three ROI protocols were 0.432、0.826、0.957.The ICC values of diffusivity from one radiologist within one month by three ROI protocols were 0.699、0.924、0.970.The ICC values of kurtosis from one radiologist within one month by three ROI protocols were 0.683、0.730、0.864.Band-Altman plots showed that the diffusivity and kurtosis measured by single slice method and whole-volume method distributed more concentrated than small solid samples method.It showed that diffusivity and kurtosis measured by single slice method and whole-volume method were more consistent and reproducible than measured by small solid samples method.Conclusions : Kurtosis and diffusivity measured by single slice method and whole-volume method in rectal cancer were not statistically significant different and had good consistency and reproducibility.Part two.The value of DKI derived parameters to histopathologic grades of rectal adenocarcinoma at 3.0 Tesla magnetic resonancePurpose: To explore the value of diffusion kurtosis imaging study(DKI)–derived parameters to differentiate rectal adenocarcinoma in different histopathologic grades at 3.0 Tesla magnetic resonance.Materials and Methods: Our study is a prospective study.Sixty-four patients with rectal carcinoma were involved in this prospective study.Diffusion kurtosis imaging was performed before surgery.Kurtosis and diffusivity from DKI in patients with rectal adenocarcinoma and normal rectal wall were measured by two radiologists.IBM SPSS Statistics 20 and Med Calc v13.1.2.0 are used to analysis the data.Exploring the relationship between kurtosis,diffusivity and different T grades、N grades、WHO histopathologic grades and PDC histopathologic grades.Results: In our study,6 patients were p T1,15 patients were p T2,32 patients were p T3,11 patients were p T4.No metastatic lymph nodes were found in 43 patients,whereas 13 patients had stage p N1 tumors and 8 patients had stage p N2 tumors.By using the WHO grading criteria,12 tumors were classified as G1,39 tumors were classified as G2,and 13 tumors were classified as G3.By using the PDC grading criteria,30 tumors were classified as G1,17 tumors were classified as G2,and 17 tumors were classified as G3.Parameters(kurtosis and diffusivity)from normal rectal wall and rectal adenocarcinoma were tested by using the Paired Sample T test,and are statistically significant different(t=8.115、-8.216,P<0.001)。Kurtosis and diffusivity were not statistically significant different in T stages(F= 1.834、2.591,P=0.151、0.061)by using One-way ANOVA test.Kurtosis and diffusivity were not statistically significant different in N stages(F=0.564、0.524,P=0.572、0.594)by using One-way ANOVA test.Kurtosis was statistically significant different in WHO grades(F=3.615,P=0.033)by using One-way ANOVA test.Diffusivity was not statistically significant different in WHO grades(F=2.018,P=0.143)by using One-way ANOVA test.Kurtosis was statistically significant different in G1 and G3(P=0.009)by using LSD test.Kurtosis was not statistically significant different in G1 and G2,G2 and G3(P=0.067、0.158)by using LSD test.Kurtosis and diffusivity were statistically significant different in PDC grades(F=11.085、10.105,P<0.001)by using One-way ANOVA test.Kurtosis and diffusivity were statistically significant different in G1 and G3,G2 and G3(P<0.001、P=0.003)by using LSD test.Differences in performance were analyzed by comparing the area under the receiver operating characteristic curve(AUC)。In normal rectal wall and rectal adenocarcinoma,the AUCs were 0.940、0.983 for kurtosis and diffusivity,respectively.In different PDC grades,the AUCs were 0.842、0.856 for kurtosis and diffusivity,respectively.Conclusions:Kurtosis and diffusivity may be valuable to differentiate between rectal adenocarcinoma and normal rectal wall and to differentiate rectal adenocarcinoma in different PDC grades.