Purpose: In vivo, the attenuation of diffusion-weighted imaging (DWI) signal at high b-values is sometimes nonlinear when plotted with semilogarithmic function and is fit well by a biexponential function. Previous reports have indicated that the fast and slow component fractions of the apparent diffusion coefficient (ADC) can be derived by biexponential fitting and that these fractions correspond to the actual diffusion components in the extra- and intracellular space. In this study, we investigated the clinical utility of DWI for the breast by performing DWI using multiple b-factors on healthy volunteers and clinical subjects, analyzing the signal by fitting it with a biexponential equation, and comparing the fitting parameters of breast lesions.
Patients and Methods: We investigated 8 healthy women as normal cases and 80 female patients with a total of 100 breast tumors (42 benign, 58 malignant tumors) as clinical cases. We performed DWI using 12 b-values for the healthy cases and 6 b-values for the clinical cases, up to a maximum b-value of 3500 s/mm².
Results: Decay of DWI signal of normal mammary glands, most cysts, and some fibroadenomas showed a monoexponential relationship, and conversely, that of intraductal papilloma (IDP) and malignant tumors was well fitted by a biexponential function. Comparison of parameters derived from biexponential fitting demonstrated no significant difference between benign and malignant lesions. For malignant tumor subtype, the fast component fraction of noninvasive ductal carcinoma was statistically greater than that of invasive ductal carcinoma.
Conclusions: Although the parameters from biexponential fitting may reflect the character of tumor cellularity, because pathological diagnosis was performed with an emphasis on cell configuration or shape rather than cellularity, it was difficult to distinguish malignant from benign tumors, including many IDPs, or to distinguish tissue types using DWI signal attenuation alone.