Biomedical Signal Processing and System Identification

我研究的长期目标是开发和验证用于生物医学信号和系统的定量分析的通用工具集。这里提出的工作将建立两个特定领域的工具：关节僵硬和心肺信号。动态关节神经力学的系统辨识方法关节的神经力学行为源于关节和肌肉的固有特性以及肌肉的反射激活。神经力学在控制姿势和运动方面很重要，前者决定了对干扰的反应，后者定义了必须控制的内部负荷。动态关节刚度将神经力学定量地描述为关节位置和扭矩之间的关系。目前，在位置和扭矩几乎保持不变的姿势条件下，对关节刚度有很好的理解。然而，在功能性活动中，当位置和扭矩有大的、快速的变化时，人们对关节僵硬知之甚少。这项工作的目标是开发工具来测量和模拟功能活动执行过程中的动态关节刚度。这对于了解正常运动功能的潜在机制和伴随运动系统疾病的变化是至关重要的。这些工具的使用将允许对关节力学进行定量、客观的评估，以诊断和监测病理变化，评估治疗方式，并在知情的情况下设计和控制假肢、矫形器和康复设备。心脏呼吸信号的信号处理与分类呼吸和心脏信号的分析对于包括睡眠呼吸暂停在内的许多疾病的诊断和监测都很重要。目前的“黄金标准”分析方法是手动对心肺信号进行评分，以确定呼吸暂停等重大事件。这种方法是非常劳动密集型的，而且受制于得分手之间和得分者内部的巨大变异性。这项工作的目标是开发一套自动化工具，将呼吸信号分类为临床识别的模式。由此产生的工具将有广泛的应用，包括：对睡眠障碍呼吸的评估，对包括早产儿呼吸暂停在内的儿童呼吸暂停的管理，婴儿从机械通气中拔出，以及非侵入性支持方式的评估。这项研究产生的算法将为研究人员提供解决神经肌肉控制和心肺功能方面的重要问题的手段。为了促进将这些工具转化为实践，我们将把它们作为全面的、随时可用的工具箱提供，并附上大量的文件、样本数据集、教程和演示。我们还将举办研讨会，向研究人员、学生和工业从业者传授如何使用我们的工具箱来满足他们的研发需求

The long term objectives of my research are to develop and validate a general purpose tool-set for the quantitative analysis of biomedical signals and systems. The work proposed here will build tools for two specific areas: joint stiffness and cardio-respiratory signals. System Identification Methods for Dynamic Joint Neuromechanics A joint's neuromechanical behavior arises from the intrinsic joint and muscle properties combined with reflex activation of muscle. Neuromechanics is important in the control of posture, where it determines the response to disturbances, and movement, where it defines the internal load that must be controlled. Dynamic joint stiffness describes neuromechanics quantitatively as the relation between joint position and torque. Currently, there is a good understanding of joint stiffness under postural conditions, where position and torque remain nearly constant. However, little is known about joint stiffness during functional activities, when there are large, rapid changes in position and torque. The objective of this work is to develop tools to measure and model dynamic joint stiffness during the execution of functional activities. This is essential to understand the mechanisms underlying normal motor function and the changes accompanying motor system disease. Use of these tools will permit the quantitative, objective evaluation of joint mechanics for the diagnosis and monitoring of pathological changes, evaluation of treatment modalities, and the informed design and control of prosthetic, orthotic, and rehabilitation devices. Signal Processing and Classification of Cardio Respiratory Signals The analysis of respiratory and cardiac signals is important for the diagnosis and monitoring of many diseases including sleep apnea. The current "gold-standard" analysis approach is to score cardiorespiratory signals manually to identify significant events such as apneas. This approach is very labor intensive, and subject to large inter- and intra- scorer variability. The objective of this work is to develop an automated tool-set to classify respiratory signals into clinically recognized patterns. The resulting tools will have a wide range of applications including: the evaluation of sleep disordered breathing, the management of pediatric apneas including the apnea of prematurity, the extubation of infants from mechanical ventilation, and the evaluation of non-invasive support modalities. The algorithms resulting from this research will provide researchers with the means to address important questions in neuromuscular control and cardio-respiratory function. To facilitate the translation of these tools to practice, we will make them available as comprehensive, ready-to-use toolboxes, accompanied by extensive documentation, sample data sets, tutorials and demonstrations. We will also hold workshops to teach researchers, students, and industrial practitioners how to use our toolboxes to address their R&D needs