Non-Mass Lesions on Automated Breast Ultrasound: An Exploratory Analysis of Imaging Features Associated with...

Objectives: This study aimed to investigate imaging features associated with malignancy and to differentiate true non-mass lesions from transient non-mass ABUS findings on automated breast ultrasound (ABUS). Methods: This retrospective study included 133 women with non-mass lesions, each contributing a single lesion assessed as BI-RADS categories 0, 4, or 5 on ABUS between January 2020 and October 2022. We analyzed multiple ABUS imaging features, including background echotexture, distribution, microcalcification-like echogenic foci, posterior shadowing, architectural distortion on axial and coronal views, and volume change. These findings were statistically compared according to pathologic results for malignancy and ultrasonographic correlation for true versus transient non-mass ABUS findings. Results: Among the 133 lesions, 15 (11.3%) were malignant and 65 (48.9%) were transient non-mass ABUS findings. Segmental distribution (odds ratio [OR] = 40.42; 95% confidence interval [CI], 4.45-367.27; p = 0.001) and coronal architectural distortion (OR = 53.59; 95% CI, 6.26-458.90; p < 0.001) showed exploratory associations with malignancy. A non-heterogeneous background echotexture (OR = 0.07; 95% CI, 0.02-0.33; p = 0.001), lack of posterior shadowing (OR = 0.14; 95% CI, 0.03-0.60; p = 0.008), and volume change (OR = 15.79; 95% CI, 1.47-170.12; p = 0.023) were associated with true lesions. Conclusions: Segmental distribution and coronal architectural distortion were associated with malignancy on ABUS in this exploratory study. Lack of posterior shadowing and the presence of volume change were associated with true lesions, while heterogeneous background echotexture was associated with transient non-mass ABUS findings. These ABUS findings may contribute to the characterization of non-mass lesions but, given the small number of malignant lesions, require cautious interpretation and validation in larger studies.