Optimization of Magnetic Resonance Fingerprintingusing Quantum Inspired Algorithms

使用量子启发算法优化磁共振指纹识别

• 批准号: 10428464
• 负责人: Debra McGivney
• 金额: $ 20.13万
• 依托单位: CASE WESTERN RESERVE UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-06-15 至 2024-03-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10428464
• 关键词: 3-Dimensional; Abdomen; Acceleration; Algorithms; Area; Benchmarking; Brain; Breast; Categories; Characteristics; Clinic; Clinical; Computing Methodologies; Coupled; Data; Detection; Disease; Fingerprint; Goals; Image; Literature; Magnetic Resonance; Magnetic Resonance Imaging; Malignant Neoplasms; Maps; Methodology; Methods; Monitor; Myelin; Nerve Degeneration; PF4 Gene; Patients; Pattern; Phase; Physiologic pulse; Problem Solving; Process; Property; Prostate; Recovery; Research; Sampling; Scanning; Signal Transduction; Speed; Techniques; Testing; Time; Tissues; Translating; Validation; Variant; Water; brain tissue; clinical application; cost; design; disease diagnostic; experience; high dimensionality; improved; in vivo; novel; quantitative imaging; quantum; quantum computing; reconstruction; tissue mapping; volunteer

Abstract Magnetic resonance fingerprinting (MRF) is a quantitative technique that is able to produce maps of tissue property values in a single and rapid acquisition. MRF has been shown to be sensitive to subtle changes in both normal and diseased tissues in the brain, prostate, breast, and abdomen, yet increased sensitivity is desired for clinical applications. A comprehensive optimization of the MRF acquisition is required to achieve higher sensitivity and faster acquisitions. We propose to apply quantum inspired optimization (QIO) techniques to solve the problem of MRF optimization. QIO methods are effective in handling large and nonconvex problems such as this one, and we propose to apply these algorithms to optimize both sequence parameters such as flip angle, repetition time, and echo time, as well as the sampling trajectories. The objective function to be optimized will be designed to include characteristics such as signal magnitude, pattern matching metrics, and T1 and T2 errors. Sequences will be tested in phantom and in vivo, and compared to current literature on MRF optimization. Such a comprehensive optimization of MRF has not been performed, and by applying these novel computational methods, we will achieve a MRF sequence that is faster and is more sensitive to changes in tissue properties for the purposes of disease detection, characterization, and monitoring.

摘要 磁共振指纹（MRF）是一种能够产生组织图的定量技术 在一个单一的和快速的收购财产价值。磁流变液已被证明是敏感的微妙变化，在这两个 脑、前列腺、乳房和腹部中的正常和患病组织，但需要提高灵敏度， 临床应用。需要全面优化MRF采集，以实现更高的 灵敏度和更快的采集。我们建议应用量子启发优化（QIO）技术来解决 MRF优化问题。QIO方法在处理大型和非凸问题时是有效的，例如 这一个，我们建议应用这些算法来优化两个序列参数，如翻转角， 重复时间和回波时间，以及采样轨迹。待优化的目标函数将 被设计为包括诸如信号幅度、模式匹配度量以及T1和T2误差之类的特性。 序列将在体模和体内进行测试，并与当前关于MRF优化的文献进行比较。等 还没有进行MRF的全面优化，并且通过应用这些新的计算 方法，我们将实现更快，对组织特性变化更敏感的MRF序列 用于疾病检测、表征和监测的目的。