Ultrasound clutter and noise improvements applied to echocardiographic left atrial appendage visualization

总结 超声成像是美国最广泛使用的先进成像模式， 平面X线摄影，它占所有成像研究的44%。不幸的是，超声图像往往是亚光学的， 一些超声心动图研究显示高达98%的患者出现退化。这种降解速度和 临床超声的徒手性质意味着图像质量比其他图像质量更依赖于操作员技能 先进的模式，但即使是最熟练的医生，超声完全失败，在11-64%的临床 任务我们的总体目标是从根本上降低超声故障率。低质量 超声在超声心动图中尤其成问题，因为它作为诊断的一线工具， 以及监测心脏病当经胸超声心动图不足时， 比如经食管超声心动图或核磁共振成像为了（1）促进高容量的经胸回波，（2）IM- 证明患者舒适度和安全性（即减少经食管回声），（3）降低成本和手术时间，以及（4） 为了最大限度地提高医生的效率，始终保持高质量的经胸超声心动图至关重要。为了做到这一点，我们 介绍了新的超声成像采集和重建解决方案。 左心耳超声心动图成像的重要性日益增加， 房颤的发病率，目前影响多达2-6百万美国人，预计将增长 到2050年达到600 - 1200万。在房颤患者中，血栓可能出现在多达12-26%的左心室。 大部分出现在附件中的心房。因此，在当前房颤患者中， 在进行心脏复律之前，必须用超声心动图对附件进行血栓筛查， 使心脏恢复窦性心律此外，由于慢性阻塞性肺疾病而接受左心耳闭合装置的患者， 房颤需要每年进行经食道超声心动图随访以监测设备。对于这两种情况，黄金- 标准和标准护理是经食管超声心动图，而左心耳可以 经胸可视化很少使用，因为图像质量被认为太不可靠。 超声图像质量会受到杂波和热噪声的影响，因此为了进行临床变换， 超声的常规改进-特别是超声心动图-有必要解决杂波 问题和噪音问题。解决这两个问题变得越来越重要，因为 波束成形的最新发展集中在非线性算法（包括深度网络）上， 这些方法的性能与噪声水平的联系更强 方法.为此，我们介绍了一种解决方案，提高信噪比使用编码激励和 使用深度网络波束形成减少杂波的综合解决方案。

Summary Ultrasonic imaging is the most widely used advanced imaging modality in the United States, and excluding planar radiography, it represents 44% of all imaging studies. Unfortunately, ultrasound images are often subopti- mal, and some echocardiography studies show degradation in up to 98% of patients. This rate of degradation and the freehand nature of clinical ultrasound means that image quality is more dependent on operator skill than other advanced modalities, but even with the most skilled practitioners, ultrasound completely fails in 11-64% of clinical tasks. Our overall goal for this application is to fundamentally decrease ultrasound failure rates. Low-quality ultrasound is particularly problematic in echocardiography because of its role as a first-line tool for diagnosing and monitoring cardiac disease. When transthoracic echo is inadequate more costly and time-consuming exams like transesophageal echo or MRI are required. To (1) facilitate the high volumes of transthoracic echo, (2) im- prove patient comfort and safety (i.e. reduce transesophageal echo), (3) reduce cost and procedure time, and (4) maximize physician efficiency, it is crucial to have consistently high quality transthoracic echo. To provide this, we introduce new acquisition and reconstruction solutions for ultrasound imaging. Echocardiographic imaging of the left atrial appendage is growing in importance in lockstep with the rising incidence of atrial fibrillation, which currently effects as many as 2-6 million Americans and is expected to grow to 6-12 million by 2050. In patients with atrial fibrillation, thrombus may appear in as many as 12-26% of left atria with most appearing in the appendage. Therefore, in patients with current atrial fibrillation the left atrial appendage must be screened with echocardiography for thrombus before cardioversion can be performed to return the heart to sinus rhythm. Further, patients receiving a left atrial appendage closure device due to chronic atrial fibrillation require annual transesophageal echo follow ups to monitor the device. For both cases, the gold- standard and standard-of-care is transesophageal echocardiography, and while the left atrial appendage can be visualized transthoracically, is is rarely used because the image quality is considered too unreliable. Ultrasound image quality is degraded by both clutter and thermal noise, so to make clinically transforma- tional improvements to ultrasound–and echocardiography specifically–it is necessary to address both the clutter problem and the noise problem. Addressing both of these issues is becoming increasingly important because recent developments in beamforming have focused on non-linear algorithms (including deep networks), and the performance of these methods are more strongly linked to the noise level compared to conventional beamforming methods. To this end, we introduce both a solution enhancing signal-to-noise ratio using coded excitation and a comprehensive solution for clutter reduction using deep network beamforming.