数据效率高的贝叶斯学习用于辐射动态MR重建
Sherine Brahma1,2, Christoph Kolbitsch1,3, Joerg Martin1
1Physikalisch-Technische Bundesanstalt (PTB), Braunschweig and Berlin, Germany.
Medical physics
|June 27, 2023
概括
基于物理学的深度学习 (DL) 有效地量化了心脏MRI重建中的不确定性,改善了图像质量,并将文物与病理区分开来. 这种方法通过为加速动态成像提供可靠的不确定性指标来提高诊断准确性.
科学领域:
- 医疗成像医学成像
- 人工智能的人工智能
- 生物医学工程 生物医学工程
背景情况:
- 心脏MRI是心血管评估的黄金标准,但由于运动和缓慢的获取,它面临着挑战.
- 深度学习 (DL) 对MRI重建有希望,但可以引入文物.
- 量化重建不确定性对于在复杂的MRI中识别DL产生的文物至关重要.
研究的目的:
- 通过使用基于物理的DL方法量化加速2D多线圈动态辐射MRI重建中的不确定性.
- 证明基于物理的DL在减少不确定性和提高图像质量方面优于无模型DL的优势.
- 为了有效地处理大规模的,高维的MRI重建问题.
主要方法:
- 扩展了基于物理的2D U-Net (XT-YT U-Net) 用于使用蒙特卡洛脱落和高斯负日志概率损失进行不确定性量化 (UQ).
- 训练并验证了XT-YT U-Net在健康志愿者的2D动态辐射MRI数据上,并对患者数据进行了测试.
- 通过使用校准图表进行UQ评估,比较了基于物理的和模型不可知的神经网络 (NN).
主要成果:
- 与模型无关的DL相比,基于物理的DL实现了更高的图像质量 (NRMSE,PSNR,SSIM) 和更低的不确定性.
- 通过校准图表验证,XT-YT U-Net显示了更好的UQ.
- 不确定性信息有效地区分了心脏MRI中的解剖结构和文物.
结论:
- 基于物理的NN (XT-YT U-Net) 成功量化了要求2D多线圈动态MRI成像的不确定性.
- 将采购模型集成到NN架构中改善了图像质量,并减少了重建不确定性.
- UQ为评估DL重建性能和可靠性提供了宝贵的见解.
相关概念视频
One-Compartment Open Model: Wagner-Nelson and Loo Riegelman Method for ka Estimation
587
This lesson introduces two critical methods in pharmacokinetics, the Wagner-Nelson and Loo-Riegelman methods, used for estimating the absorption rate constant (ka) for drugs administered via non-intravenous routes. The Wagner-Nelson method relates ka to the plasma concentration derived from the slope of a semilog percent unabsorbed time plot. However, it is limited to drugs with one-compartment kinetics and can be impacted by factors like gastrointestinal motility or enzymatic degradation.
On...
On...
587
Magnetic Resonance Imaging
5.3K
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...
5.3K


