Histogram analysis of diffusion kurtosis imaging of nasopharyngeal carcinoma: Correlation between quantitative parameters and clinical stage

// Xiao-Quan Xu 1, * , Gao Ma 1, * , Yan-Jun Wang 1 , Hao Hu 1 , Guo-Yi Su 1 , Hai-Bin Shi 1 and Fei-Yun Wu 1 1 Department of Radiology, The First Affiliated Hospital of Nanjing Medical University, Nanjing, China * These authors contributed equally to this work Correspondence to: Fei-Yun Wu, email: wfy_njmu@163.com Keywords: nasopharyngeal carcinoma, clinical stage, magnetic resonance imaging, diffusion kurtosis imaging, histogram Received: January 17, 2017 Accepted: April 16, 2017 Published: May 03, 2017 ABSTRACT Purpose: To evaluate the correlation between histogram parameters derived from diffusion-kurtosis (DK) imaging and the clinical stage of nasopharyngeal carcinoma (NPC). Results: High T-stage (T3/4) NPC showed significantly higher K app-mean ( P = 0.018), K app-median ( P = 0.029) and K app-90th ( P = 0.003) than low T-stage (T1/2) NPC. High N-stage NPC (N2/3) showed significantly lower Dapp-mean ( P = 0.002), Dapp-median ( P = 0.002) and Dapp-10th ( P < 0.001) than low N-stage NPC (N0/1). High AJCC-stage NPC (III/IV) showed significantly lower D app-10th ( P = 0.038) than low AJCC-stage NPC (I/II). ROC analyses indicated that Kapp-90th was optimal for predicting high T-stage (AUC, 0.759; sensitivity, 0.842; specificity, 0.607), while Dapp-10th was best for predicting high N- and AJCC-stage (N-stage, AUC, 0.841; sensitivity, 0.875; specificity, 0.807; AJCC-stage, AUC, 0.671; sensitivity, 0.800; specificity, 0.588). Materials and Methods: DK imaging data of forty-seven consecutive NPC patients were retrospectively analyzed. Apparent diffusion for Gaussian distribution (D app ) and apparent kurtosis coefficient (K app ) were generated using diffusion-kurtosis model. Histogram parameters, including mean, median, 10th, 90th percentiles, skewness and kurtosis of D app and K app were calculated. Patients were divided into low and high T, N and clinical stage based on American Joint Committee on Cancer (AJCC) staging system. Differences of histogram parameters between low and high T, N and AJCC stages were compared using t test. Multiple receiver operating characteristic (ROC) curves were used to determine and compare the value of significant parameters in predicting high T, N and AJCC stage, respectively. Conclusions: DK imaging-derived parameters correlated well with clinical stage of NPC, therefore could serve as an adjunctive imaging technique for evaluating NPC.