Value of whole-lesion apparent diffusion coefficient (ADC) first-order statistics and texture features in clinical...

To explore the value of whole-lesion apparent diffusion coefficient (ADC) first-order statistics and texture features in clinical staging of cervical cancers.

Fifty-six women (mean age, 51 years) with histopathologically confirmed cervical cancers underwent 3 T pelvic magnetic resonance imaging including diffusion-weighted imaging (b=0, 800 s/mm²) prospectively. The ADC first-order statistics and texture features derived from the whole volume of cervical cancers were correlated with International Federation of Gynecology and Obstetrics (FIGO) stages (i.e., stages I, II, III, and IV).

The first-order statistics of skewness, kurtosis, and entropy, and the texture features of entropy(H) and homogeneity correlated positively, while the texture feature of energy correlated negatively with FIGO stages (all p<0.05). Skewness, kurtosis, entropy, entropy(H), and homogeneity were significantly higher, while energy was significantly lower in cervical cancers at higher (IIB-IVA) than lower (IB-IIA) FIGO stages (all p<0.05). Kurtosis and energy had the largest areas under the receiver operating characteristic (ROC) curve of 0.749 and 0.746 in differentiating cervical cancers at lower (IB-IIA) from higher (IIB-IVA) FIGO stages.

Whole-lesion ADC first-order statistics and texture features proved relevant and meaningful in the clinical staging of cervical cancers.