# Impact of contrast-enhanced ultrasound optimized breast imaging reporting and data system category on the biopsy decision for non-mass breast lesions

## TL;DR

This study shows that using contrast-enhanced ultrasound improves accuracy in deciding whether to biopsy non-mass breast lesions.

## Contribution

A CEUS-optimized BI-RADS model significantly improves diagnostic accuracy and biopsy decision-making for non-mass breast lesions.

## Key findings

- CEUS features like crab claw-like enhancement and enlarged enhancement extent are independent risk factors for malignancy.
- CEUS optimization increased diagnostic accuracy from 58.1% to 87.8% and reduced biopsy rates while improving malignancy detection.
- Using BI-RADS 4A or 4B as biopsy thresholds showed improved performance after CEUS optimization.

## Abstract

To investigate the impact of contrast-enhanced ultrasound (CEUS) optimized Breast Imaging Reporting and Data System (BI-RADS) prediction model on the biopsy decision for non-mass breast lesions (NMLs).

148 NMLs histopathologically confirmed from 142 patients who underwent ultrasound (US) and CEUS examination were retrospectively enrolled. US and CEUS features were compared between malignant and benign NMLs. A CEUS-optimized BI-RADS category prediction model was developed using logistic regression. The diagnostic performance and impact on biopsy decision of US and CEUS-optimized BI-RADS category were compared.

Of 148 NMLs, 77 were malignant and 71 benign. CEUS features such as earlier wash-in, later wash-out, hyperenhancement, heterogeneous enhancement, enlarged enhancement extent, and crab claw-like enhancement were significantly associated with malignancy (P<0.05). The enlarged enhancement extent and crab claw-like enhancement were independent risk factors for malignancy (P<0.05). Using BI-RADS 4B as the cutoff, the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy for diagnosing NMLs increased from 53.2%, 63.4%, 61.2%, 55.6%, and 58.1% before optimization to 93.5%, 81.7%, 84.7%, 92.1%, and 87.8% after CEUS optimization, respectively. With BI-RADS 4A or 4B as the biopsy threshold, the biopsy rate, malignancy detection rate, and missed diagnosis rate were 100%, 52.03%, and 0% before optimization, changing to 75%, 66.67%, and 3.90% after optimization for 4A, respectively. The corresponding values for 4B were 45.95%, 60.29%, and 46.75% before optimization, and 57.43%, 84.71%, and 6.49% after optimization, respectively.

The CEUS-optimized BI-RADS category prediction model could provide valuable guidance for biopsy decision-making in NMLs.