EAGER: Design of an Active Voice Box Prosthesis with Embedded Actuation

晚期喉癌（通常称为声带癌）最有效的治疗方法通常是切除整个器官，即，“全喉切除术”“声带的丧失，通常称为声带，它被安置在喉部，导致声带所能实现的所有三个关键生理功能的严重并发症：呼吸，吞咽和发声。 呼吸和吞咽已经受到相当大的关注，喉切除术的患者，但发声仍然提出了未解决的问题，由于声带动力学的复杂性和可变性。 目前的喉切除术治疗方案采用不自然的语音产生方法，并需要通过重定向气道呼吸。 该项目通过调查创建一个人工语音箱的可行性来解决这个限制，该人工语音箱可以重新创建动态行为或取代健康声带组织的功能。基于对猪声带特性的研究，开发了一种3D打印技术来生产合成声带，可以微调以具有所需的运动特性。 振动将通过在打印的声带中嵌入压电致动器（可以将能量转化为运动的小型电机）来产生。 将开发一种微创闭环控制系统，以产生适当的信号并控制合成声带行为。 还将努力使用来自健康受试者的EEG（脑电图）（即脑信号）远程控制人工语音盒，以估计受试者移动声带的意图。 拟议的研究提供了改善喉切除术患者生活质量的可能性，这些患者可能会因无法沟通而产生深刻的孤立感。EEG驱动的鲁棒控制策略的发展也将为指导其他损伤的康复策略提供一个框架。获得的结果将通过会议和期刊论文传播，并通过专门的网站，如Connexions和国家科学数字图书馆提供。这个探索性项目的目标是研究开发一种具有嵌入式驱动的人工语音盒的可行性，该语音盒通过EEG使用生物反馈，并利用新颖的自适应控制算法来恢复需要喉切除术的患者的全部发声功能。与目前采用非自然语音产生方法并需要通过重定向气道呼吸的喉切除术治疗方案不同，这种方法专注于创建，控制和动态测试人工语音盒，该人工语音盒可以重建动态行为或取代健康声带组织的功能。初步工作的重点是确定猪声带的连续模型参数，以及生产合成声带所需的增材制造技术，这是该项目的基础。 研究计划分为六项任务。 任务1）动态测试，以评估切除的猪声带和合成声带模型的声带振动。当样品动态加载时，将记录振动频率。以每秒3000帧的速度捕获的数字图像将能够确定样品的上级表面上的应变场和变形。任务2）比较和迭代设计合成声带，以了解加载样本时基频如何变化。 合成声带模型的空间特性将通过调谐制造工艺参数来迭代地调整，以便在动态测试期间实现期望的频率响应。 PZT（压电）驱动将嵌入到打印的声带中，目的是在动态测试期间调节声带振动。任务3）通过开发基于输出的控制算法来控制声带，而无需人工参与，该算法将解决诸如时间延迟、致动器幅度和速率饱和限制以及部分和噪声测量等影响。任务4）使用来自健康受试者的EEG信号控制声带，目的是改变当恒定的空气流穿过合成声带时产生的声音的音高。 任务5）对任务3和任务4中收集的数据进行数据后处理和性能评估。将比较使用标准控制器和基于输出的控制器获得的闭环结果。一个性能指标，基于实施的容易性，调整和闭环系统的实际性能将被开发和使用，以评估与现有的algorithm.This奖项反映了NSF的法定使命的控制性能，并已被认为是值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估的支持。

The most effective treatment for advanced cancer of the larynx, commonly called the voice box, is often the removal of the entire organ, i.e., a "total laryngectomy." Loss of the vocal folds, commonly called vocal cords, which were housed in the larynx, leads to significant complications in all three of the critical physiological functions that the vocal folds enabled: respiration, swallowing, and voice production. Breathing and swallowing have received considerable attention for patients with laryngectomy, but vocalization still presents unresolved issues due to the complexity and variability of the vocal fold dynamics. Current laryngectomy treatment options employ unnatural speech production methods and require breathing through a redirected airway. This project addresses this limitation by investigating the feasibility of creating an artificial voice box that can recreate the dynamic behavior or replace the functionality of healthy vocal fold tissue. Based on studies of the properties of pig vocal folds, a 3D printing technique has been developed to produce synthetic vocal folds that can be fine-tuned to have desired movement properties. Vibrations will be created by embedding piezoelectric actuators (small motors that can convert energy into motion) in the printed vocal folds. A minimally invasive, closed-loop control system will be developed to generate appropriate signals and control the synthetic vocal fold behavior. Efforts will also be made to control the artificial voice box remotely using EEG (electroencephalogram), i.e. brain signals, from a healthy subject to estimate the subject's intent to move the vocal folds. The proposed research offers the possibility of improving the quality of life of laryngectomy patients who could develop a profound sense of isolation from their inability to communicate. Development of an EEG-driven robust control strategy will also provide a framework for guiding rehabilitation strategies for other impairments. Results obtained will be disseminated through conference and journal papers and be made available using dedicated websites such as Connexions and the National Science Digital Library. The goal of this exploratory project is to investigate the feasibility of developing an artificial voice-box, with embedded actuation, that uses biofeedback via EEG and leverages novel adaptive control algorithms to restore full vocal function to patients requiring a laryngectomy. Unlike current laryngectomy treatment options that employ unnatural speech production methods and require breathing through a redirected airway, this approach focuses on creating, controlling, and dynamically testing an artificial voice box that can recreate the dynamic behavior or replace the functionality of healthy vocal fold tissue. Preliminary work focused on identifying continuous model parameters for porcine vocal folds and the additive manufacturing techniques necessary to produce synthetic vocal folds serves as the foundation for the project. The Research Plan is organized under six tasks. TASK 1) Dynamic testing to evaluate vocal fold vibration of both excised porcine vocal folds and the synthetic vocal fold models. Frequencies of vibration will be recorded as the samples are dynamically loaded. Digital images captured at 3000 frames per second will enable the determination of strain fields and deformation on the superior surface of the samples. TASK 2) Comparison and iterative design of synthetic vocal fold in order to understand how the fundamental frequencies change while the samples are loaded. The spatial properties of the synthetic vocal fold model will be iteratively adjusted by tuning manufacturing process parameters in order to achieve a desired frequency response during dynamic tests. PZT (piezoelectric) actuation will be embedded into the printed vocal folds with the goal of modulating the vocal fold vibration during dynamic testing. TASK 3) Controlling vocal folds without a human in the Loop through development of an Output Based Control algorithm that will address effects such as time delays, actuator amplitude and rate saturation limitations, and partial and noisy measurements. TASK 4) Controlling vocal folds using EEG signals from a healthy subject, for the purpose of changing the pitch of the sound produced when a constant flow of air is passed across the synthetic vocal folds. TASK 5) Data post-processing and performance evaluation of the data collected in Tasks 3 and 4. The closed loop results obtained using standard controllers and the Output Based controller will be compared. A performance index, based on the ease of implementation, tuning and actual performance of the closed loop system will be developed and used to assess the performance of the control with respect to existing algorithms.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.