Lightweight U-Net for High-Resolution Breast Imaging

TL;DR

This paper introduces a lightweight U-Net architecture designed for high-resolution breast imaging, balancing segmentation accuracy with computational efficiency for breast cancer screening.

Contribution

It proposes a simplified U-Net model optimized for high-resolution breast images, improving efficiency without sacrificing significant accuracy.

Findings

Achieved effective segmentation with reduced computational load.

Demonstrated potential for real-time breast cancer screening applications.

Balanced model complexity and performance in high-resolution imaging.

Abstract

We study the fully convolutional neural networks in the context of malignancy detection for breast cancer screening. We work on a supervised segmentation task looking for an acceptable compromise between the precision of the network and the computational complexity.