Characterization of rectal cancer: Whole-tumour histogram analysis of diffusion kurtosis imaging at 3 T

Abstract Purpose: This study aimed to explore the performance of whole-tumour histogram analysis of diffusion kurtosis imaging (DKI) in the characterization of rectal cancer.Methods: Forty-two rectal cancer patients were retrospectively enrolled. Histogram analysis of D, K and the conventional apparent diffusion coefficient (ADC) were calculated. Student’s t test or the Mann–Whitney U test and receiver operating characteristic (ROC) curves were used for statistical analysis.Results: The ADCSD and ADCkurtosis were significantly higher in N (+) patients than in N0 patients (p<0.05). The ADCSD, DSD and Kkurtosis were significantly higher in PNI-positive patients than in PNI-negative patients (p<0.05). Dskewness was significantly higher in KRAS mutation-type patients than in KRAS wild-type patients (p<0.05). The area under the curve (AUC) of the multiparametric ADCSD and ADCkurtosis values for the detection of metastatic lymph nodes was 0.816 (p<0.001); the sensitivity and specificity were 81.0% and 76.2%, respectively. The AUC of the multiparametric ADCSD, DSD, and Kkurtosis values for the detection of PNI was 0.830 (p<0.001); the sensitivity and specificity were 80% and 88.9%, respectively. Dskewness can be used to determine the KRAS status; the cut-off value, AUC, sensitivity and specificity were 0.465, 0.732 (p=0.010), 78.3%, and 73.7%, respectively. Conclusion: Whole-tumour histogram analysis of DKI could be used to identify both pathological and genetic characteristics of rectal cancer.