Collaborative Research: SCH: Quantifying Cardiac Performance by Measuring Myofiber Strain with Routine MRI

合作研究：SCH：通过常规 MRI 测量肌纤维应变来量化心脏性能

• 批准号: 2205043
• 负责人: Luigi Perotti
• 金额: $ 69.63万
• 依托单位: The University of Central Florida Board of Trustees
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-09-01 至 2026-08-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2205043&HistoricalAwards=false
• 关键词: Collaborative; Research; SCH; Quantifying; Cardiac

The goal of this research is to develop a new method to quantify cardiac performance in patients affected by cardiac diseases. Current strategies to evaluate cardiac performance often rely on inadequate global measures, such as ejection fraction, which are non-specific and often late outcomes. Cardiac motion is driven by billions of heart cells acting together, whose contraction and relaxation can be measured using myofiber strain. Myofiber strain is therefore a direct measure of cardiac function and is an ideal candidate to evaluate cardiac performance, improving diagnosis and therapy planning. However, there are three main obstacles that hinder the deployment of myofiber strain in a clinical setting: (i) There is no method to reliably compute myofiber strain from images that are routinely acquired; (ii) There are no reliable error estimates for the evaluated strains, preventing their use to distinguish between health and disease; and (iii) There is no framework to compute myofiber strain on demand without hardware and technical barriers. This project aims at overcoming these obstacles by combining computational modeling and artificial intelligence with readily available magnetic resonance imaging. The transition to the clinic will be highly facilitated by deploying the proposed framework in a completely online platform leveraging on-demand cloud computing. Investigators around the globe will be able to test remotely the newly proposed technology without the need for specific hardware or additional software. The multidisciplinary research carried out in this project will train the next generation of scientists, who will be capable of carrying out projects in smart health and biomedical research at the forefront of medical imaging, artificial intelligence, and computational modeling. The proposed approach will estimate myofiber strain by minimizing the difference between computed and measured surface cardiac motion. Measured surface motion is extracted from cine Magnetic Resonance Imaging (MRI), which is routinely acquired in a clinical MRI setting. Computed left ventricular surface motion is obtained by solving a computational kinematics model based on the biomechanics of myofiber shortening and relaxation. Uncertainty in myofiber strain predictions will be evaluated based on imaging data noise and model assumptions. Fast and accurate high-fidelity models and Bayesian error estimators will propagate experimental and model uncertainties to establish confidence in myofiber strain estimates. As a results, the generated models will allow to characterize strains’ uncertainty and variation in healthy and diseased individuals. The proposed approach will be demonstrated and validated in a pilot study to aid therapy planning in patients affected by aortic stenosis. This new approach paves the way to improve diagnosis, prognosis, and therapy planning for patients affected by a wide range of cardiomyopathies resulting in compromised left ventricular function and therefore myofiber mechanics.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

本研究的目的是开发一种新的方法来量化心脏疾病患者的心脏功能。目前评估心脏性能的策略通常依赖于不充分的整体测量，例如射血分数，这是非特异性的并且通常是晚期结果。心脏运动由数十亿心脏细胞共同作用驱动，其收缩和舒张可以使用肌纤维应变测量。因此，肌纤维应变是心脏功能的直接量度，并且是评价心脏性能、改进诊断和治疗计划的理想候选者。 然而，有三个主要障碍阻碍了肌纤维应变在临床环境中的部署：（i）没有从常规获取的图像可靠地计算肌纤维应变的方法;（ii）对于所评估的应变没有可靠的误差估计，阻止了它们用于区分健康和疾病;和（iii）没有在没有硬件和技术障碍的情况下按需计算肌纤维应变的框架。该项目旨在通过将计算建模和人工智能与现成的磁共振成像相结合来克服这些障碍。通过在利用按需云计算的完全在线平台中部署拟议框架，将大大促进向诊所的过渡。地球仪的研究人员将能够远程测试新提出的技术，而不需要特定的硬件或额外的软件。在该项目中进行的多学科研究将培养下一代科学家，他们将能够在医学成像，人工智能和计算建模的最前沿开展智能健康和生物医学研究项目。所提出的方法将通过最小化计算的和测量的表面心脏运动之间的差异来估计肌纤维应变。从电影磁共振成像（MRI）中提取测量的表面运动，这是在临床MRI环境中常规采集的。计算的左心室表面运动是通过求解基于肌纤维缩短和松弛的生物力学的计算运动学模型来获得的。将根据成像数据噪声和模型假设评估肌纤维应变预测的不确定性。快速准确的高保真模型和贝叶斯误差估计器将传播实验和模型的不确定性，以建立肌纤维应变估计的信心。因此，生成的模型将允许表征健康和患病个体中的菌株的不确定性和变化。所提出的方法将在一项试点研究中得到证明和验证，以帮助受主动脉瓣狭窄影响的患者制定治疗计划。这一新方法为改善因各种心肌病导致左心室功能受损和肌纤维力学受损的患者的诊断、预后和治疗计划铺平了道路。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Ventricular Helix Angle Trends and Long-Range Connectivity

心室螺旋角趋势和远距离连通性

• 发表时间: 2023
• 期刊: Lecture notes in computer science
• 作者: Wilson, Alexander J.;Han, Q. Joyce;Perotti, Luigi E.;Ennis, Daniel B.
• 通讯作者: Ennis, Daniel B.

Anatomically-guided deep learning for left ventricle geometry generation with uncertainty quantification based on short-axis MR images

基于短轴 MR 图像的解剖学引导深度学习，通过不确定性量化生成左心室几何形状

• DOI: 10.1016/j.engappai.2023.106012
• 发表时间: 2023
• 期刊: Engineering Applications of Artificial Intelligence
• 影响因子: 8
• 作者: Zuben, Andre Von;Perotti, Luigi E.;Viana, Felipe A.C.
• 通讯作者: Viana, Felipe A.C.

Long Axis Cardiac MRI Segmentation Using Anatomically-Guided UNets and Transfer Learning

使用解剖引导 UNet 和迁移学习进行长轴心脏 MRI 分割

• 发表时间: 2023
• 期刊: Lecture notes in computer science
• 作者: Von Zuben, Andre;Whitt, Emily;Viana, Felipe A.C.;Perotti, Luigi E.
• 通讯作者: Perotti, Luigi E.