General Linear Modeling For Magnetic Resonance Spectroscopy

磁共振波谱学的一般线性建模

• 批准号: 10509724
• 负责人: Georg Oeltzschner
• 金额: $ 22.03万
• 依托单位: JOHNS HOPKINS UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-07-05 至 2025-03-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10509724
• 关键词: Address; Adoption; Algorithms; Behavior; Biochemical; Brain; Cloud Computing; Code; Communities; Computer Models; Concentration measurement; Coupled; Data; Diffuse; Diffusion; Dimensions; Elements; Equipment; Event; Fingerprint; Freedom; Grant; High Performance Computing; Human; Individual; Internet; Knowledge; Linear Models; Link; Magnetic Resonance Spectroscopy; Measures; Metabolic; Methods; Modeling; Modernization; Participant; Pathology; Pharmacology; Phase; Physiologic pulse; Physiological; Property; Recovery; Relaxation; Resolution; Running; Series; Signal Transduction; Spectrum Analysis; Stimulus; Techniques; Testing; Time; Tissues; Translating; biophysical properties; candidate marker; cell type; chemical property; cloud based; density; design; diffusion weighted; experimental study; flexibility; improved; in vivo; in vivo magnetic resonance spectroscopy; innovation; model design; neurochemistry; novel; open source; physical property; potential biomarker; response; tool; user-friendly

Project Summary Advanced multi-spectrum magnetic resonance spectroscopy (MRS) methods allow the non-invasive measurement of the concentration of neurochemicals, but also of other biophysical properties. The currently available one-dimensional modeling tools cannot adequately model such data because they are incapable of incorporating prior knowledge about the relationships between sub-spectra into a single multi-dimensional model. Additional parameters that can be encoded in the acquisition, but not adequately accommodated within the quantification model include metabolite relaxation times, metabolite diffusion tensors, and physiological metabolic response to external stimulation. This project addresses the gap in currently available modeling tools for MRS by introducing a generalized linear combination modeling framework for MRS. This avoids the overfitting that arises from serial application of current one-dimensional models, dramatically increasing model parsimony. All code developed will be made available to the community open-source, and the modeling framework will be made available in the cloud via a web user interface.

项目概要 先进的多谱磁共振波谱 (MRS) 方法允许非侵入性 测量神经化学物质的浓度，还可以测量其他生物物理特性。目前的 可用的一维建模工具无法充分建模此类数据，因为它们无法 将有关子光谱之间关系的先验知识纳入单个多维模型中。 可以在采集中编码但未充分容纳在采集中的附加参数 量化模型包括代谢物弛豫时间、代谢物扩散张量和生理学 对外界刺激的代谢反应。 该项目通过引入广义线性模型来解决当前可用的 MRS 建模工具的空白 MRS 的组合建模框架。这避免了电流串行应用引起的过度拟合 一维模型，显着提高模型简约性。所有开发的代码都将公开 向社区开源，建模框架将通过网络用户在云端提供 界面。