First step for computer assisted evaluation of qualatitive supersonic shear wave elastography characteristics in breast tissue.

The addition of shear wave elastography (SWE) imaging to standard B-mode ultrasound assessment supports the differ-ential diagnosis of solid breast lesions. One way to assess SWE images is qualitative evaluation of the stiffness pattern according to the Tozaki classification. Until now this qualitative assessment has only been possible manually and so has low reproducibility and high inter-observer variability. Thus, this work aims to investigate the feasibility of automatic assessment of Tozaki categories. SWE images of 41 solid breast lesions (9 benign, 32 malignant) were assessed manually by a radiologist and automatically. The results were compared to the histology of the lesions. The diagnostic performance of the automatic assessment is similar to the manual assessment. The algorithm introduced in this work demonstrates the ability to automatically assess qualitative SWE features, given segmented lesions, according to the Tozaki classification.