针对MRI图像非均匀噪声，提出分块专家路由的稀疏混合模型。
Sparse Mixture-of-Experts for Non-Uniform Noise Reduction in MRI Images
- 按图像块或分割区域分组，基于特征相似性分配专用去噪网络
- 在合成与真实脑部MRI数据上优于现有最优方法
- 对未见数据集泛化能力强，适合临床医学图像处理
磁共振成像（MRI）是临床诊断的重要工具，但成像过程中引入的噪声伪影常影响其应用。有效去噪对提升图像质量并保留解剖结构至关重要。然而，传统方法通常假设噪声分布均匀，难以应对MRI中常见的非均匀噪声。本文在先前多分支MRI去噪方法基础上，提出一种细粒度稀疏混合专家框架。该方法将图像分解为基于块或分割的区域，根据学习到的特征相似性进行分组，并将每个区域路由至专用的卷积神经网络进行去噪。实验表明，该方法在合成与真实脑部MRI数据集上均显著优于当前最优去噪技术。此外，其在未见数据集上也表现出良好泛化能力，凸显了鲁棒性与适应性。
原文摘要 · Abstract (English)
Magnetic Resonance Imaging (MRI) is an essential diagnostic tool in clinical settings, but its utility is often hindered by noise artifacts introduced during the imaging process. Effective denoising is critical for enhancing image quality while preserving anatomical structures. However, traditional denoising methods, which often assume uniform noise distributions, struggle to handle the non-uniform noise commonly present in MRI images. Building on prior multi-branch MRI denoising approaches, we introduce a fine-grained sparse mixture-of-experts framework for MRI image denoising. Our method decomposes each image into patch-based or segmentation-based regions, groups regions according to their learned feature similarity, and routes each region to a specialized denoising convolutional neural network. Our method demonstrates superior performance over state-of-the-art denoising techniques on both synthetic and real-world brain MRI datasets. Furthermore, we show that it generalizes effectively to unseen datasets, highlighting its robustness and adaptability.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。