A comprehensive deep learning framework for MRI reconstruction

用于 MRI 重建的综合深度学习框架

• 批准号: 10211757
• 负责人: Rizwan Ahmad
• 金额: $ 61.38万
• 依托单位: OHIO STATE UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-07-01 至 2025-03-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10211757
• 关键词: 3-Dimensional; 4D MRI; Acceleration; Address; Adoption; Adult; Algorithms; Anesthesia procedures; Architecture; Awareness; Brain; Brain Neoplasms; Brain imaging; Breathing; Cardiac; Child; Childhood; Clinical; Data; Development; Diagnosis; Diagnostic; Disease; Evaluation; Exhibits; Formulation; Goals; Headache; Heart; Heart Diseases; Image; Imaging Device; Imaging technology; Immune; Investigation; Lead; Magnetic Resonance Imaging; Maps; Methods; Modeling; Motion; Network-based; Patients; Pattern; Performance; Phase; Physics; Play; Predisposition; Process; Pythons; Recovery; Rest; Sampling; Scanning; Sedation procedure; Speed; Structure; Techniques; Technology; Testing; Time; Training; Variant; Work; base; cardiovascular imaging; computer framework; convolutional neural network; cost; data acquisition; data space; deep learning; denoising; design; heart imaging; high dimensionality; image reconstruction; imaging modality; improved; learning strategy; musculoskeletal imaging; neural network architecture; novel; pediatric patients; prospective; real-time images; reconstruction; repository

PROJECT SUMMARY/ABSTRACT The primary goal of this investigation is to develop and validate a comprehensive, robust deep learning (DL) framework that improves MRI reconstruction beyond the limits of existing technology. The proposed framework uses “plug-and-play” algorithms to combine physics-driven MR acquisition models with state-of-the-art learned image models, which are instantiated by image denoising subroutines. To fully exploit the rich structure of MR images, we propose to use DL-based denoisers that are trained in an application-specific manner. The proposed framework, termed PnP-DL, offers advantages over other existing DL methods, as well as compressed sensing (CS). Compared to existing DL methods for MRI reconstruction, PnP-DL is more immune to inevitable variations in the forward model, such as changes in the coil sensitivities or undersampling pattern, allowing it to generalize across applications and acquisition settings. Compared to CS, PnP-DL recovers images faster, with higher quality, and with potentially superior diagnostic value. Our preliminary results highlight the potential of PnP-DL to advance MRI technology. In this work, we will fur- ther develop PnP-DL and validate it in these major applications: cardiac cine, 2D brain, and 3D brain imaging. In Aim 1, we will train and optimize convolutional neural network-based application-specific denoisers for the above-mentioned applications. The denoiser with the best denoising performance will be selected for further investigation. In Aim 2, we will develop and compare different PnP algorithms. The algorithm yielding the best combination of reconstruction accuracy and computational speed will be implemented in Gadgetron for inline processing. In Aim 3, we will compare the performance of PnP-DL to other state-of-the-art methods using retro- spectively undersampled data. This study will demonstrate that, in terms of image quality, PnP-DL is superior to CS and existing DL methods and, despite higher acceleration, is non-inferior to parallel MRI with rate-2 acceler- ation. In Aim 4, we will evaluate the performance of PnP-DL using prospectively undersampled data from adult and pediatric patients. Successful completion of this project will demonstrate that PnP-DL outperforms state- of-the-art methods in terms of image quality while exhibiting a level of robustness and broad applicability that has eluded other DL-based MRI reconstruction methods. The acceleration and image quality improvement afforded by these developments will benefit almost all MRI applications, including pediatric imaging, where reducing sedation is a pressing need, and high-dimensional imaging applications (e.g., whole-heart 4D flow imaging), which are too slow for routine clinical use.

项目概要/摘要 这项调查的主要目标是开发和验证全面、强大的深度学习 (DL) 改进 MRI 重建的框架超越了现有技术的限制。拟议的框架 使用“即插即用”算法将物理驱动的 MR 采集模型与最先进的学习相结合 图像模型，由图像去噪子例程实例化。充分利用MR的丰富结构 对于图像，我们建议使用基于深度学习的降噪器，并以特定于应用的方式进行训练。拟议的 称为 PnP-DL 的框架比其他现有的 DL 方法以及压缩感知具有优势 （CS）。与现有的 MRI 重建 DL 方法相比，PnP-DL 更容易受到不可避免的变化的影响 在前向模型中，例如线圈灵敏度或欠采样模式的变化，使其能够泛化 跨应用程序和采集设置。与 CS 相比，PnP-DL 恢复图像速度更快，质量更高， 并具有潜在的优越诊断价值。 我们的初步结果凸显了 PnP-DL 推进 MRI 技术的潜力。在这项工作中，我们将进一步 他们开发了 PnP-DL 并在以下主要应用中对其进行验证：心脏电影、2D 大脑和 3D 大脑成像。 在目标 1 中，我们将训练和优化基于卷积神经网络的特定于应用的降噪器 上述应用。将选择去噪性能最好的去噪器进行进一步的降噪 调查。在目标 2 中，我们将开发并比较不同的 PnP 算法。算法产生最好的结果 重建精度和计算速度的结合将在 Gadgetron 中实现内联 加工。在目标 3 中，我们将使用复古方法将 PnP-DL 的性能与其他最先进的方法进行比较 分别欠采样数据。本研究将证明，就图像质量而言，PnP-DL 优于 CS 和现有的 DL 方法，尽管加速度更高，但并不劣于具有速率 2 加速器的并行 MRI 化。在目标 4 中，我们将使用成人的前瞻性欠采样数据来评估 PnP-DL 的性能 和儿科患者。该项目的成功完成将证明 PnP-DL 的性能优于 State- 在图像质量方面采用最先进的方法，同时表现出一定程度的稳健性和广泛的适用性 已经避开了其他基于深度学习的 MRI 重建方法。加速和图像质量改善 这些发展带来的成果将有利于几乎所有 MRI 应用，包括儿科成像，其中 减少镇静是迫切需要，高维成像应用（例如，全心 4D 血流） 成像），这对于常规临床使用来说太慢了。