Qualitative Assessment of Signal Intensity Evolution Across Three b-Values in DWIBS for Breast Lesion Differentiation

To evaluate the diagnostic accuracy and inter-observer reliability of qualitative signal intensity (SI) changes across three b-values in diffusion-weighted imaging with background suppression (DWIBS) for differentiating benign from malignant breast lesions.

This prospective diagnostic accuracy study enrolled 113 consecutive female patients (mean age 44.7 ± 12.1 years) with sono-mammographically suspicious lesions who underwent breast MRI including T2-weighted imaging and DWIBS at b-values of 800, 1000, and 1500 s/mm². Three expert breast-imaging radiologists independently assessed T2WI, SI behavior across b-values, and assigned a final ACR BI-RADS MRI category. Diagnostic performance was computed using the histopathological examination as the reference standard. Chi-square/Fisher exact tests evaluated the association between SI pattern on DWIBS and malignant outcome. Inter-observer agreement was assessed with Fleiss' kappa (SI classification).

Of 113 lesions, 64 (56.6%) were malignant and 49 (43.4%) were benign. SI changes from b = 800 to 1500s/mm² yielded sensitivity of 85.7-90.5%, specificity of 100%, and accuracy of 92.0-94.7% between reader 1 and 2 considering the increased SI as the optimal cutoff for predicting malignancy. While reader 3 recorded a stable/isointense SI as the best cutoff for predicting malignancy with sensitivity of 100%, specificity of 87.5% and accuracy of 94.7%. Inter-observer agreement was almost perfect for SI changes (κ = 0.856, p < 0.001).

DWIBS across three b-values provides an alternative, contrast-free method for breast lesion characterization with high reproducibility. DWIBS merits prospective validation within full DCE-MRI protocols in larger multicenter studies before clinical adoption.