GL-UNet:A Deep Learning Model of Breast Tumor Lesion Segmentation for MRI Images

Deep learning-based image segmentation methods are widely applied in the field of medical imaging. However, due to certain quality issues such as high noise and artifacts of the breast MRI images, the performances of machine learning models could be significantly affected. To address this problem, we propose a modified nnU-Net segmentation network named GL-UNet for breast cancer MRI image segmentation. GL-UNet preprocesses the images using Gaussian-Laplacian combined filtering to extract the edges of the lesion area, resulting in a more accurate image segmentation. Compared to the original nnU-Net and several state-of-the-art networks on the breast cancer dataset, our proposed method achieves a better performance, revealing the effectiveness of GL-UNet.