Ultra-Fast High-Resolution Multi-Parametric MRI for Characterizing Cartilage Extracellular Matrix

用于表征软骨细胞外基质的超快速高分辨率多参数 MRI

• 批准号: 10929242
• 负责人: Fang Liu
• 金额: $ 64.32万
• 依托单位: MASSACHUSETTS GENERAL HOSPITAL
• 依托单位国家: 美国
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-09-21 至 2024-08-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10929242
• 关键词: Acceleration; Affect; Binding; Biochemical; Biological Markers; Cartilage; Clinical; Clinical Research; Collagen; Collagen Fiber; Communities; Degenerative polyarthritis; Disease; Evolution; Extracellular Matrix; Face; Histologic; Human; Image; Imaging Techniques; Incidence; Individual; Joints; Knee; Knee Osteoarthritis; Knee joint; Magic; Magnetic Resonance; Magnetic Resonance Imaging; Magnetic Resonance Spectroscopy; Maps; Measures; Methods; Patients; Physics; Predisposition; Proteoglycan; Protocols documentation; Relaxation; Research; Resolution; Scanning; Signal Transduction; Slice; Specificity; Statistical Methods; Structure; Techniques; Thick; Thinness; Time; Tissue Model; Tissues; Water; articular cartilage; cartilage degradation; deep learning; human subject; image reconstruction; imaging biomarker; imaging modality; in vivo; learning strategy; magnetic resonance imaging biomarker; mechanical properties; millimeter; non-invasive monitor; novel; reconstruction; research study; specific biomarkers; success; supervised learning; Acceleration; Affect; Binding; Biochemical; Biological Markers; Cartilage; Clinical; Clinical Research; Collagen; Collagen Fiber; Communities; Degenerative polyarthritis; Disease; Evolution; Extracellular Matrix; Face; Histologic; Human; Image; Imaging Techniques; Incidence; Individual; Joints; Knee; Knee Osteoarthritis; Knee joint; Magic; Magnetic Resonance; Magnetic Resonance Imaging; Magnetic Resonance Spectroscopy; Maps; Measures; Methods; Patients; Physics; Predisposition; Proteoglycan; Protocols documentation; Relaxation; Research; Resolution; Scanning; Signal Transduction; Slice; Specificity; Statistical Methods; Structure; Techniques; Thick; Thinness; Time; Tissue Model; Tissues; Water; articular cartilage; cartilage degradation; deep learning; human subject; image reconstruction; imaging biomarker; imaging modality; in vivo; learning strategy; magnetic resonance imaging biomarker; mechanical properties; millimeter; non-invasive monitor; novel; reconstruction; research study; specific biomarkers; success; supervised learning

PROJECT SUMMARY Osteoarthritis (OA) is one of the most prevalent diseases affecting human joints, characterized by decreased proteoglycan content and disruption of the collagen fiber network in the cartilage extracellular matrix. Quantitative magnetic resonance (MR) imaging has been used to quantify cartilage composition and microstructure changes due to extracellular matrix degeneration in OA research studies. While many quantitative MR techniques have been explored, existing methods face serious limitations, including lack of specificity to assess individual macromolecular components, sensitive to magic angle effect, susceptible to partial volume effect due to thick image slice. More importantly, quantitative MR techniques typically require a much longer scan time than standard imaging due to the need for repeated scans of the same imaging object at varying imaging parameters. Spatial resolution and imaging volume coverage must be compromised to make a clinically feasible scan in OA research studies. This proposal aims to develop a new imaging technique that can provide robust, sensitive, and specific imaging biomarkers for simultaneously assessing cartilage proteoglycan and collagen components, and meanwhile can be acquired at the submillimeter spatial resolution, thin image slice, and full knee coverage within a 10-min scan time. Among all the quantitative MR techniques, multi-component T2 relaxation imaging has been found to provide sensitive and specific information for cartilage proteoglycan content; cross-relaxation imaging has been found to provide complementary information regarding the collagen fiber network of cartilage. The proposal will develop a simultaneous multi-component T2 relaxation and cross-relaxation imaging technique that can provide sensitive and specific imaging biomarkers to assess proteoglycan and collagen content and their ultra-structures in a unified imaging framework (Aim 1). This imaging protocol will be optimized using rigorous statistical methods and accelerated through a novel deep learning method that leverages self-supervised learning and MR physics-informed tissue modeling. The derived MR imaging biomarkers will be correlated with tissue histological, biochemical, and mechanical properties, which will create a basis for interpretation of the clinical study results (Aim 2). A pilot clinical study using the optimized and accelerated imaging technique will be performed on patients with varying degrees of knee OA, establishing the clinical evidence of the utility, efficiency, and overall clinical value of this newly proposed technique on detecting OA incidence and predicting OA progression (Aim 3). Our proposed new methods will root from developing novel rapid image acquisition, combined with advanced deep learning reconstruction and automatic processing, all of which are pioneered by our research team. Successful completion of the proposal will provide the OA research community with a new set of MR biomarkers to non-invasively monitor disease-related and treatment-related changes in cartilage composition and ultra-structure in human subjects.

项目摘要 骨关节炎（OA）是影响人类关节的最普遍的疾病之一，其特征是 软骨外基质中胶原蛋白纤维网络的蛋白聚糖含量和破坏。定量 磁共振（MR）成像已用于量化软骨组成和微观结构的变化 在OA研究中，由于细胞外基质变性。虽然许多定量的MR技术具有 经过探索，现有方法面临严重的局限性，包括缺乏评估个人的特异性 大分子成分，对魔法角效应敏感，易受部分体积效应的影响 图像切片。更重要的是，定量MR技术通常需要比 由于需要在不同的成像参数上重复扫描相同成像对象，因此标准成像。 必须妥协空间分辨率和成像量覆盖范围，以在OA中进行临床可行的扫描 研究。该建议旨在开发一种新的成像技术，可以提供强大，敏感和 用于同时评估软骨蛋白聚糖和胶原蛋白成分的特定成像生物标志物，以及 同时，可以在亚毫米空间分辨率，薄图像切片和全膝盖覆盖范围内获取 10分钟的扫描时间。在所有定量MR技术中，多组分T2松弛成像已经 发现为软骨蛋白聚糖含量提供敏感和特定的信息；交叉解释成像 已经发现可以提供有关软骨软化软骨网络的互补信息。这 提案将开发同时多组分的T2松弛和交叉删除成像技术 可以提供敏感的特定成像生物标志物来评估蛋白聚糖和胶原蛋白含量及其 统一成像框架中的超结构（AIM 1）。该成像协议将使用严格的 统计方法，并通过一种新的深度学习方法加速，该方法利用自我监督 学习和MR物理知识的组织建模。派生的MR成像生物标志物将与 组织组织学，生化和机械性能，这将为解释的基础 临床研究结果（AIM 2）。使用优化和加速成像技术的试点临床研究将是 对具有不同程度的膝盖OA的患者进行，建立了效用，效率的临床证据 以及这项新提出的有关检测OA发病率和预测OA的技术的总体临床价值 进展（目标3）。我们提出的新方法将植根于开发新型快速图像采集， 结合先进的深度学习重建和自动处理，所有这些都是由 我们的研究团队。该提案的成功完成将为OA研究社区提供新的 一组MR生物标志物，非侵入性监测软骨的疾病相关和与治疗相关的变化 人类受试者的组成和超结构。