Study Of The Application Of Diffusional Kurtosis Model Magnetic Resonance Imaging In Diagnosis Of Rectal Cancer

PURPOSE: To evaluate the diagnosis value of Diffusional Kurtosis (DK) imaging inrectal cancer.MATERIALS AND METHODS:128rectal patients underwent normal magneticresonance and DKI (5b values0,500,1000,1500,2000;3diffusion directions) scansfrom2013.7to2014.2,28of them were operated in2weeks. Parameters maps wereobtained used permeability processing tool by Philips healthcare for the metric MK, ameasure of excess kurtosis of diffusion, and corrected diffusion (MD), a correcteddiffusion coefficient accounting for the non-Gaussian behavior, of both tumour andnormal rectal wall. All findings were analyzed by SPSS19.0。RESULTS: There were20moderately differentiated adenocarcinomas, one poorlydifferentiated adenocarcinomas, three mucinous cell carcinoma, one adenoma andthree rectitis. MD and MK values for tumour were as follows:1.30±0.27×10-3mm2/sec,1.07±0.16, MD and MK values for inflammation were1.63±0.07×10-3mm2/sec，whereas MD and MK for normal rectal walls （NRW） were1.81±0.35×10-3mm2/sec,0.83±0.21. There was a significant difference between thetumour and NRW， between adenocarcinomas and inflammation, and betweenmoderately differentiated adenocarcinomas (MDAC) and mucinous cellcarcinoma(MCC)， respectively（p<0.05）. MD and MK had a large area under thereceiver operating characteristics（ROC） curve for predicting rectal cancer (ROC,0.907and0.849)。The sensitivity and specificity for discriminating between cancerand NRW were found for MD (92%and79%, respectively) and MK(83%and68%,respectively). Optimal thresholds for MD and MK for differentiating cancer fromNRW were≤1.446×10-3mm2/sec and≥0.908, respectively. MD and MK alsohad a large area under the receiver operating characteristics（ROC） curve forpredicting MDAC(ROC,0.857and0.889)。 The sensitivity and specificity fordiscriminating between MDAC and MCC were found for MD (100%and57%,respectively) and MK(86%and67%, respectively). Optimal thresholds for MD and MK for differentiating MDAC from MCC were≤1.263×10-3mm2/sec and≥0.958, respectively.DK parameters did not significantly differ between normal rectalwall and inflammation，or patients with and without lymphatic involvement, or indifferent pTNM staging and Dukes staging（p>0.05）. DK parameters revealed nosignificant correlation with T staging, N staging，pathologic classification, Dukesstaging（p>0.05）CONCLUSION: Preliminary findings suggest that DK imaging is able to helpdifferentiate among normal rectal wall, inflammation, adenocarcinomas and is able todepict microstructural changes within rectal cancer.