Application of the diffusion kurtosis model for the study of breast lesions

Abstract

Objectives: To evaluate diffusion-weighted imaging (DWI) and diffusion kurtosis imaging (DKI) in the differentiation and characterisation of breast lesions.

Methods: Thirty-six women underwent breast magnetic resonance imaging (MRI) including a DWI sequence with multiple b-values (50-3,000 s/mm(2)). Mean values for apparent diffusion coefficient (ADC), mean diffusivity (MD) and mean kurtosis (MK) were calculated by lesion type and histological subtype. Differences and correlation between parameters were determined.

Results: Forty-four lesions were found. There were significant differences between benign and malignant lesions for all parameters (ADC, p = 0.017; MD, p = 0.028; MK, p = 0.017). ADC and MD were higher for benign (1.96 ± 0.41 × 10(-3) and 2.17 ± 0.42 × 10(-3) mm(2)/s, respectively) than for malignant lesions (1.33 ± 0.18 × 10(-3) and 1.52 ± 0.50 × 10(-3) mm(2)/s). MK was higher for malignant (0.61 ± 0.27) than benign lesions (0.37 ± 0.18). We found differences between invasive ductal carcinoma (IDC) and fibroadenoma (FA) for all parameters (ADC, MD and MK): p = 0.016, 0.022 and 0.016, respectively. FA and fibrocystic change (FC) showed differences only in MK (p = 0.016).

Conclusions: Diffusion in breast lesions follows a non-Gaussian distribution. MK enables differentiation and characterisation of breast lesions, providing new insights into microstructural complexity. To confirm these results, further investigation in a broader sample should be performed.