Clinical Report: Enhanced Detection of Breast Lesions in Ultrasound Imaging
Overview
This study presents a framework for breast lesion detection in ultrasound images, utilizing a Dual-Stream Mamba Aggregation (DSMA) module and a Structure-aware Axial Attention (SAA) module.
Background
Breast cancer is a leading cause of cancer diagnoses among women, making early detection crucial for effective treatment. Ultrasound imaging is commonly used for breast evaluation but faces challenges such as noise and low contrast, which can hinder accurate lesion detection.
Data Highlights
Parameter
Value
Parameters
2.50M
GFLOPs
6.4
FPS
161.29
Key Findings
The proposed framework integrates DSMA and SAA modules for improved feature representation.
Competitive detection performance was achieved on the BUV and WH-BUS datasets.
The method maintains a lightweight structure with only 2.50M parameters.
Robustness and visualization analyses demonstrated the complementary benefits of the proposed modules.
Detection performance is enhanced while keeping computational overhead low.
Clinical Implications
This framework addresses limitations of existing methods in breast ultrasound imaging.
Conclusion
The study introduces a framework that models contextual and structural features for breast lesion detection in ultrasound images.