Whole-lesion histogram analysis of apparent diffusion coefficient for distinguishing cervical cancers with different differentiation

The purpose of our study was to evaluate the value of whole-lesion apparent diffusion coefficient (ADC) histogram parameters in distinguishing cervical cancers with different differentiation. A cohort of 18 patients (mean age, 53 years) with biopsy confirmed cervical cancers were involved in this prospective study. All of them underwent 3-T diffusion weighted imaging (DWI) with b values of 0 and 800 s/mm2. ADC histogram parameters were obtained from the entire tumor, including mean ADC values (ADCmean), the percentile ADC values i.e. ADC5%, ADC10%, ADC25%, ADC50%, ADC75%, and ADC90%, skewness, kurtosis, s-sDlowest and s-sDav distribution width. According to Mann-Whitney U-test, the s-sDav and 90th percentile ADC values of well or moderately differentiated cervical cancers were significantly higher than those of poorly differentiated cervical cancers (Both P < 0.05). Whereas the other measurements showed no significant differences. The ROC analysis demonstrated that the area under the ROC (AUC) of s-sDav (0.925) was higher than that of 90th percentile ADC values (0.85). All parameters showed excellent performance in inter-observer stability assessment (all ICC > 0.95). In conclusion, whole-lesion histogram analysis of ADC maps is helpful in discriminating well or moderately differentiated cervical cancers from poorly differentiated cervical cancers.