SNRAware:通过信号对噪声比单位训练和G因子地图增强改进了深度学习MRI否定
Hui Xue1, Sarah M Hooper2, Iain Pierce3
1Health Futures, Microsoft Research, 14820 NE 36th St, Bldg 99, Rm 4941, Redmond, WA 98052.
Radiology. Artificial intelligence
|October 22, 2025
概括
一种新的深度学习方法,SNRAware,通过利用图像重建的定量噪声数据来增强MRI无声化. 这种方法可以改善各种MRI应用中的图像质量和模型概括性.
科学领域:
- 医疗成像医学成像
- 人工智能的人工智能
- 放射学 放射学是一门学科.
背景情况:
- 磁共振成像 (MRI) 对于诊断至关重要,但容易受到噪音的影响,这可能会降低图像质量.
- 深度学习方法已经显示出对MRI解密的希望,但性能可能受到概括问题的限制.
研究的目的:
- 开发和评估基于深度学习的MRI无声化方法 (SNRAware),该方法包含从图像重建中获得的定量噪声分布信息.
- 提高MRI无噪声模型的性能和概括能力.
主要方法:
- 一项回顾性研究使用了大量数据集 (来自 96,605 个心脏电影系列的 2,885,236 张图像),采用 3-T MRI 扫描仪获取.
- 该SNRAware培训计划模拟了各种数据集,并使用了MRI重建的定量噪声分布数据.
- 评估了14个模型架构 (基于卷积和变压器),重点是3D输入张量和架构不可知论.
主要成果:
- 在内部和外部测试数据集上,SNRAware显著提高了MRI无效性能,超过了没有重建知识的训练模型.
- 与卷积模型相比,变压器模型表现出更高的性能,3D输入张量器比2D图像产生了更好的结果.
- 性能最好的模型在不同的MRI序列 (实时cine, perfusion,脑,脊柱) 和场强度 (1.5-T和3-T) 中得到了很好的概括,显示出明显的对比度和噪声比率改善.
结论:
- 该SNRAware培训计划有效地利用重建数据进行基于深度学习的MRI解密,提高性能和概括性.
- 这种方法为改善各种MRI应用中的图像质量提供了强大的解决方案.
相关概念视频
Magnetic Resonance Imaging
9.0K
Magnetic resonance imaging (MRI) is a noninvasive medical imaging technique based on a phenomenon of nuclear physics discovered in the 1930s, in which matter exposed to magnetic fields and radio waves was found to emit radio signals. In 1970, a physician and researcher named Raymond Damadian noticed that malignant (cancerous) tissue gave off different signals than normal body tissue. He applied for a patent for the first MRI scanning device in clinical use by the early 1980s. The early MRI...
9.0K
Imaging Studies IV: Magnetic Resonance Imaging
221
Introduction:Magnetic Resonance Imaging, or MRI, can include a specialized imaging technique of the urinary system known as Magnetic Resonance Urography (MRU). This radiation-free technique uses strong magnetic fields and radio waves to produce detailed images with the help of a computer. MRU is particularly effective for visualizing fluid-filled structures like the kidneys, ureters, and bladder.Applications of MRI in the Genitourinary SystemKidneys and Ureters: MRI detects tumors, cysts,...
221


