Controlling Quality and Capturing Uncertainty in Advanced Diffusion Weighted MRI

控制质量并捕捉高级扩散加权 MRI 的不确定性

• 批准号: 10683306
• 负责人: Bennett A. Landman
• 金额: $ 63.25万
• 依托单位: VANDERBILT UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2015
• 资助国家: 美国
• 起止时间: 2015-09-20 至 2025-06-30
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10683306
• 关键词: Affect; Aging; Alzheimer' s Disease; Alzheimer' s disease diagnosis; Alzheimer' s disease related dementia; Alzheimer’s disease biomarker; American; Anatomy; Architecture; Biological Markers; Blood Vessels; Brain; Brain Mapping; Calibration; Cerebrum; Chemicals; Clinical Research; Clinical Sciences; Clinical Trials; Cognitive; Coupled; Data; Data Analyses; Data Set; Databases; Detection; Diffusion Magnetic Resonance Imaging; Disease; Disease Progression; Dissection; Early Diagnosis; Early Intervention; Educational workshop; Ensure; Evaluation; FDA approved; Functional disorder; Goals; Healthcare; Histology; Image; Individual; Intervention; Investigation; Machine Learning; Maps; Measures; Modality; Modeling; Morphologic artifacts; Neurons; Noise; Normal Range; Patient Care; Patients; Pharmaceutical Preparations; Physiological; Prognosis; Property; Prospective cohort; Protocols documentation; Public Health; Quality Control; Research; Resources; Role; Scanning; Sensitivity and Specificity; Severity of illness; Site; Software Tools; Specificity; Structure; System; Techniques; Tissue Model; Tissues; Translating; Uncertainty; Variant; Visualization software; aging brain; biomarker development; clinical application; cohort; data integration; deep neural network; disease prognostic; effective intervention; imaging modality; imaging study; improved; in vivo; innovation; magnetic resonance imaging biomarker; neuroimaging; novel therapeutics; open source; prospective; success; tool; tractography; translational impact; virtual; virtual biopsy; white matter; white matter change

PROJECT SUMMARY Alzheimer’s Disease and related dementia are a growing public health crisis affecting 5.8 million Americans, yet there are only four FDA-approved medications for Alzheimer’s Disease, none of which are disease-modifying. Hence, early detection and diagnosis are key to successful patient management and biomarkers are needed for evaluating new therapies in clinical trials. White matter changes are increasingly implicated in early Alzheimer’s Disease progression, and diffusion weighted magnetic resonance imaging (DW-MRI) has been included in many national-scale studies. Yet, quantitative investigation of DW-MRI data is hindered by a lack of consistency due to variation in acquisition protocols, sites, and scanners. DW-MRI enables quantification of brain microstructure and facilitates structural connectivity mapping. Substantial recent progress has been made with calibration and harmonization to reduce inter-subject variance and improve interpretability of computed measures. Yet, the fundamental challenge remains that clinical application of DW-MRI (as currently implemented) is confounded by inter-scanner and inter-site effects. To improve understanding of structural changes in Alzheimer’s Disease, we will construct and evaluate three separate analysis strategies to characterize, calibrate, and optimize DW-MRI for single-subject biomarker development for Alzheimer’s Disease. We will integrate and optimize our strategies using large retrospective multi-site studies and validate the approaches on two distinct prospective cohorts. Specifically, we aim to: Aim 1: Optimize data-driven techniques for stability across sessions, scanners/sites, and field strengths Impact: Harmonized DW-MRI methods will increase sensitivity to Alzheimer’s Disease and its prodromal stages. Aim 2: Translate innovations in microstructural harmonization to structural connectivity (tractography) Impact: Harmonizing structural connectivity will improve understanding of white matter in Alzheimer’s Disease. Aim 3: Advance statistical tools for single-subject inference through normative database construction Impact: Data-driven resources for uncertainty estimation will enable robust single-single subject inference. Relevance and Impact on Healthcare: The proposed research will advance understanding of Alzheimer’s Disease through (1) quantitative harmonization of DW-MRI biomarkers, (2) protocols for harmonization of retrospective and prospective DW-MRI studies, and (3) new tools for single subject inference targeting older cohorts. We will organize workshops/challenges to maximize the translational impact on clinical science. The long-term goal of our research is to (1) provide a well-validated strategy to quantitatively evaluate DW-MRI data across sites, (2) enhance DW-MRI biomarkers for Alzheimer’s Disease, and (3) advance patient care. Our research strategy will transform the manner in which DW-MRI data are interpreted and enable single-subject machine learning to interpret brain properties. The resources, software, and visualization tools will be made freely available in open source through DIPY to facilitate continued innovation.

项目概要 阿尔茨海默病和相关痴呆症是一场日益严重的公共卫生危机，影响着 580 万美国人，但 FDA 批准的治疗阿尔茨海默病的药物只有四种，其中没有一种能够缓解疾病。 因此，早期检测和诊断是成功患者管理的关键，并且需要生物标志物 在临床试验中评估新疗法越来越多地与早期阿尔茨海默病有关。 疾病进展和扩散加权磁共振成像 (DW-MRI) 已被纳入许多研究中 然而，由于缺乏一致性，DW-MRI 数据的定量研究受到阻碍。 DW-MRI 能够根据采集协议、部位和扫描仪的变化来量化大脑微观结构。 并促进了结构连通性测绘，最近在校准和测量方面取得了实质性进展。 协调以减少主体间差异并提高计算测量的可解释性。 根本挑战仍然是 DW-MRI（目前实施的）的临床应用 被扫描仪间和站点间的影响所混淆。 为了加深对阿尔茨海默病结构变化的理解，我们将构建并评估三个 单独的分析策略来表征、校准和优化单受试者生物标志物的 DW-MRI 我们将通过大型回顾来整合和优化我们的策略。 具体来说，我们的目标是： 目标 1：优化数据驱动技术，以实现跨会话、扫描仪/站点和场强的稳定性 影响：统一的 DW-MRI 方法将提高对阿尔茨海默病及其前驱阶段的敏感性。 目标 2：将微观结构协调的创新转化为结构连通性（纤维束成像） 影响：协调结构连接将提高对阿尔茨海默病白质的理解。 目标 3：通过规范的数据库建设，推进单主体推理的统计工具 影响：用于不确定性估计的数据驱动资源将实现稳健的单一主题推理。 对医疗保健的相关性和影响：拟议的研究将增进对阿尔茨海默病的了解 通过 (1) DW-MRI 生物标志物的定量协调，(2) 协调一致的方案来治疗疾病 回顾性和前瞻性 DW-MRI 研究，以及 (3) 针对老年人的单受试者推理新工具 我们将组织研讨会/挑战，以最大限度地提高对临床科学的转化影响。 我们研究的长期目标是 (1) 提供一种经过充分验证的策略来定量评估 DW-MRI 数据 跨站点，(2) 增强阿尔茨海默病的 DW-MRI 生物标志物，(3) 促进患者护理。 研究策略将改变 DW-MRI 数据的解释方式并实现单一受试者 将制作用于解释大脑特性的机器学习资源、软件和可视化工具。 通过 DIPY 以开源方式免费提供，以促进持续创新。

项目成果

Harmonizing 1.5T/3T Diffusion Weighted MRI through Development of Deep Learning Stabilized Microarchitecture Estimators.

通过开发深度学习稳定微架构估计器来协调 1.5T/3T 扩散加权 MRI。

• DOI: 10.1117/12.2512902
• 发表时间: 2019
• 期刊: Proceedings of SPIE--the International Society for Optical Engineering
• 作者: Nath,Vishwesh;Remedios,Samuel;Parvathaneni,Prasanna;Hansen,ColinB;Bayrak,RozaG;Bermudez,Camilo;Blaber,JustinA;Schilling,KurtG;Janve,VaibhavA;Gao,Yurui;Huo,Yuankai;Lyu,Ilwoo;Williams,Owen;Resnick,Susan;Beason-Held,Lori;Ro
• 通讯作者: Ro

Reproducibility Evaluation of SLANT Whole Brain Segmentation Across Clinical Magnetic Resonance Imaging Protocols.

跨临床磁共振成像协议的 SLANT 全脑分割的再现性评估。

• DOI: 10.1117/12.2512561
• 发表时间: 2019
• 期刊: Proceedings of SPIE--the International Society for Optical Engineering
• 作者: Xiong,Yunxi;Huo,Yuankai;Wang,Jiachen;Davis,LTaylor;McHugo,Maureen;Landman,BennettA
• 通讯作者: Landman,BennettA

Federated Gradient Averaging for Multi-Site Training with Momentum-Based Optimizers.

• DOI: 10.1007/978-3-030-60548-3_17
• 发表时间: 2020-10
• 期刊: Lecture notes-monograph series
• 作者: Remedios SW;Butman JA;Landman BA;Pham DL
• 通讯作者: Pham DL

Distributed deep learning for robust multi-site segmentation of CT imaging after traumatic brain injury.

分布式深度学习，用于脑外伤后 CT 成像的稳健多部位分割。

• DOI: 10.1117/12.2511997
• 发表时间: 2019
• 期刊: Proceedings of SPIE--the International Society for Optical Engineering
• 作者: Remedios,Samuel;Roy,Snehashis;Blaber,Justin;Bermudez,Camilo;Nath,Vishwesh;Patel,MayurB;Butman,JohnA;Landman,BennettA;Pham,DzungL
• 通讯作者: Pham,DzungL

Communication-efficient federated learning for multi-institutional medical image classification

• DOI: 10.1117/12.2611654
• 发表时间: 2022-04
• 作者: Shuang Zhou;B. Landman;Yuankai Huo;A. Gokhale