The Study of a 231 French Patient Cohort Significantly Extends the Mutational Spectrum of the Two Major Usher Genes MYO7A and USH2A

Usher syndrome is an autosomal recessive disorder characterized by congenital hearing loss combined with retinitis pigmentosa, and in some cases, vestibular areflexia. Three clinical subtypes are distinguished, and MYO7A and USH2A represent the two major causal genes involved in Usher type I, the most severe form, and type II, the most frequent form, respectively. Massively parallel sequencing was performed on a cohort of patients in the context of a molecular diagnosis to confirm clinical suspicion of Usher syndrome. We report here 231 pathogenic MYO7A and USH2A genotypes identified in 73 Usher type I and 158 Usher type II patients. Furthermore, we present the ACMG classification of the variants, which comprise all types. Among them, 68 have not been previously reported in the literature, including 12 missense and 16 splice variants. We also report a new deep intronic variant in USH2A. Despite the important number of molecular studies published on these two genes, we show that during the course of routine genetic diagnosis, undescribed variants continue to be identified at a high rate. This is particularly pertinent in the current era, where therapeutic strategies based on DNA or RNA technologies are being developed.     

基于深度神经网络的sEMG手势识别研究

为了提高表面肌电信号（sEMG）手势识别算法的准确性,并解决人为提取大量特征具有局限性的问题,提出了一种基于深度神经网络的手势识别方法。将MYO臂环采集到的8通道sEMG数据,采用活动段分割的方法探测到有效动作;设计出一种融合卷积神经网络（CNN）和长短时记忆（LSTM）网络的神经网络;实验的结果表明手势识别准确率为91.6%,验证了提出的方案高效可行。     

基于MYO和Android的肌电假肢手控制系统设计与实现*

针对目前假肢手控制系统成本高、操作不灵活、实用性差等问题,设计了一种基于MYO和Android的肌电假肢手控制系统。在Android平台上,开发了一款智能终端应用,实现了对MYO臂环采集的表面肌电信号进行数据处理和模式识别,并实时控制假肢手完成五种模式。实验结果表明,系统的在线识别率可达98.2%,并可在300ms左右完成一次识别过程,满足了假肢手控制的精度和实时性要求。该系统设计成本低廉、方便携带且易于扩展,很好地满足了截肢患者对假肢手控制的需求。     

聚苯乙烯磁性微球的羧基化修饰

采用新颖的油-水界面法合成了油溶性Fe3O4纳米粒子,同时运用改进的微乳液聚合法分别对Fe3O4纳米粒子和苯乙烯（St）进行乳化,形成甲基丙烯酸甲酯（MMA）修饰的聚苯乙烯磁性纳米复合微球,通过红外光谱（FF-IR）、透射电镜（TEM）、X射线衍射（XRD）、振动样品磁强计（VSM）进行表征,探讨st与MMA质量比、乳化剂浓度和Fe3O4含量对所形成的复合微求性能的影响,最后,通过抗原抗体的结合反应的吸光度变化测试,表明小鼠抗人肌红蛋白抗体（MYO）可成功连接在复合微球表面,从而为实现其体外诊断应用打下基础。     

基于表面肌电和加速度信息融合的手势识别

针对现有手势识别方法中人为提取特征具有局限性的问题,考虑到MYO臂环稀疏多通道的特点,提出一种融合表肌电信号(sEMG)和加速度(ACC)信息的手势识别方案。首先由5位受试者佩戴MYO臂环,同步采集8组不同手势的多通道sEMG、ACC数据;其次采用一种新的数据预处理方法,使用高能量滑动窗将数据进行分割,得到短时有效活动段;最后设计出一种并行多层的LSTM网络,对两类数据进行特征提取和融合,实验准确率达96.87%。结果表明,所提出的手势识别方案简便可行。     

基于组合RNN网络的EMG信号手势识别

肌肉计算机接口(MCI)系统是虚拟现实、人机交互研究的热点之一,其核心问题是EMG肌电信号分类,因而MCI系统可以与深度学习方法有效结合。表面EMG信号分为高密度瞬时信号与稀疏多通道信号,前者类似于图像,可以采用CNN网络处理;本文应用RNN网络对后者进行研究,并利用MYO臂环实现了相应MCI系统。稀疏多通道EMG信号是不定长时间序列信号,前后时间相关性高,采用RNN网络进行分类。通过对原始信号进行时域、时频域、频域特征拓展,获得原始信号的多流特征序列,并提出两类组合RNN网络架构处理相应多流信号。用户依赖时算法准确率达90.78%,非用户依赖的人群测试中手势识别准确率达78.01%,实时动作识别准确率达82.09%,算法能在61.7毫秒内识别手势动作。本文所提出的组合RNN网络方法可以有效区分基于EMG信号的不同动作,且所设计的MCI系统用户泛化性与工作实时性表现好。     

Functional Characterization of the MYO6 Variant p.E60Q in Non-Syndromic Hearing Loss Patients

Hereditary hearing loss (HHL) is a common genetic disorder accounting for at least 60% of pre-lingual deafness in children, of which 70% is inherited in an autosomal recessive pattern. The long tradition of consanguinity among the Qatari population has increased the prevalence of HHL, which negatively impacts the quality of life. Here, we functionally validated the pathogenicity of the c.178G>C, p.E60Q mutation in the MYO6 gene, which was detected previously in a Qatari HHL family, using cellular and animal models. In vitro analysis was conducted in HeLa cells transiently transfected with plasmids carrying MYO6^WT or MYO6^p.E60Q, and a zebrafish model was generated to characterize the in vivo phenotype. Cells transfected with MYO6^WT showed higher expression of MYO6 in the plasma membrane and increased ATPase activity. Modeling the human MYO6 variants in zebrafish resulted in severe otic defects. At 72 h post-injection, MYO6^p.E60Q embryos demonstrated alterations in the sizes of the saccule and utricle. Additionally, zebrafish with MYO6^p.E60Q displayed super-coiled and bent hair bundles in otic hair cells when compared to control and MYO6^WT embryos. In conclusion, our cellular and animal models add support to the in silico prediction that the p.E60Q missense variant is pathogenic and damaging to the protein. Since the c.178G>C MYO6 variant has a 0.5% allele frequency in the Qatari population, about 400 times higher than in other populations, it could contribute to explaining the high prevalence of hearing impairment in Qatar.     

基于人工神经网络对sEMG信号的手势动作识别

针对利用表面肌电信号(sEMG)对手势动作的肌电信号的研究较少和sEMG信号处理过于复杂的问题,提出了利用人工神经网络和sEMG信号对人的手势动作进行识别研究,引入了MYO硬件设备对新的手势动作sEMG信号采集.利用MYO从手臂上获取每一个手势动作的sEMG信号,提取信号特征值,作为算法的训练数据和测试数据.采用人工神经网络中的反向传递神经网络算法来进行对4种不同手势动作分类,对应目标手指识别率在90.35%.研究结果可以被用来做临床诊断和生物医学的应用以及用于现代硬件的发展和更现代化的人机交互的发展.     

面向肌电信号的虚拟现实提线木偶动画研究

泉州提线木偶属于首批中国非物质文化遗产，是中华传统文化的实现形式之一.然而，由于木偶体积庞大，携带与操作不便，直接限制了其受众群体.为了实现提线木偶的有效传承与保护，设计了基于手势识别的虚拟现实提线木偶动画方案，构建了基于MYO臂环肌电信号的人体生理信号控制动画原型系统，应用两个用户实验验证了算法的高精确性与易操控性.首先，通过低通滤波与平滑实现多通道肌电信号数据的信号处理.其次，提取八通道时域特征与时频域特征，并通过线性判别器将其降维为六维特征向量，实现特征间关联性消除与算法鲁棒性增强.最后，构造多分类支撑向量机实现特征向量，确定手势识别结果.实验验证算法离线动作平均识别准确率为95.59%，实时动作平均识别准确率达到90.75%，在1.1s左右完成手势识别.面向提线木偶任务，构造了两个用户体验任务，普通用户人群中，木偶动作识别率较高，用户使用意愿、易学性等方面，系统性能亦显著高于真实木偶操控；专业用户在承认系统可用性的同时，具有较高的接受度.用户任务表明该设计满足了手势识别实时性和准确性的要求，具有良好的交互性和趣味性.相关研究可以广泛地应用于计算机动画等类似的系统，对于体验和保护提线木偶具有现实意义.