SBIR Phase II: Compact, Low-cost, Automated 3D Ultrasound System for Regular and Accessible Breast Imaging

SBIR二期项目的更广泛/商业影响引入了乳腺癌筛查的新模式，提供了一个具有成本效益和可访问的个性化乳房健康监测平台，使妇女及其医生能够获得准确和可操作的数据。在美国，每年有超过30万女性被诊断出乳腺癌，4万女性死于乳腺癌。如果早期发现，乳腺癌的存活率为99%，但目前筛查的成本、敏感性和可及性方面的限制导致三分之一的癌症在早期阶段被遗漏。早期发现与较低的治疗成本有关，每年可节省数十亿美元的直接医疗保健和生产力损失，这表明更好的乳腺癌筛查平台具有明显的经济和社会效益。该技术充分利用了自动化超声波的优势和基于云计算的人工智能的新功能，扩展了这些系统的部署，包括低资源环境，如无预约或农村诊所、药房，以及在家中进行自我监测。该平台的便携性、低成本、2分钟扫描时间、自动分析和以患者为中心的设计极大地提高了乳腺癌筛查的可及性和采用率，从而获得更好的临床结果，并减少了美国医疗保健系统的成本负担。SBIR二期项目建议继续开发一种新型平台，该平台将3D自动超声与人工智能（AI）相结合，以实现个性化和可访问的乳房成像。拟议的项目将提高紧凑型扫描仪和可穿戴配件组合的性能，以产生独立于操作员培训的可重复图像；这可以在2分钟内完成，不需要昂贵的设备、电离辐射或患者不适。直观的软件将使医生可视化整个乳房体积和准确定位和测量病变。人工智能将识别异常肿块，并预测恶性肿瘤的可能性，帮助医生准确快速地诊断。第二阶段的研发重点是五个目标：(i)优化系统性能，通过新的波束形成技术实现更高分辨率、更高帧率和更快的扫描，从而实现高质量的全乳房成像；（ii）最终确定可穿戴和扫描仪的设计，以水为耦合介质，确保可靠运行；（iii）就临床使用的安全性和功能要求进行可用性验证和确认；（iv）进行小型研究，以验证扫描仪发现现有乳房病变的能力；(v)开发一种机器学习引擎，用于实时检测和表征超声扫描仪获取的图像中的病变。该奖项反映了美国国家科学基金会的法定使命，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。

The broader/commercial impact of this SBIR Phase II project introduces a new paradigm in breast cancer screening with a cost-effective and accessible platform for personalized breast health monitoring, empowering women and their physicians with accurate and actionable data. In the US, over 300,000 women are diagnosed and 40,000 women die from breast cancer annually. Breast cancer has a 99% survival rate if detected early, but limitations in cost, sensitivity, and accessibility of current screening result in missing 1 in 3 cancers at early stages. Early detection is associated with lower costs of treatment that save billions of dollars in direct medical care and lost productivity annually, demonstrating a clear economic and societal benefit for better breast cancer screening platforms. The technology leverages the proven benefits of automated ultrasound and the newfound power of cloud-based artificial intelligence to expand the deployment of these systems, including lower-resource settings such as walk-in or rural clinics, pharmacies, and in the home for self-monitoring. The platform's portability, low cost, 2-minute scan time, automated analysis, and patient-centered design greatly increases the accessibility and adoption of breast cancer screening, resulting in better clinical outcomes and a reduced cost burden to the US healthcare system. This SBIR Phase II project proposes to continue development of a novel platform that combines 3D automated ultrasound with artificial intelligence (AI) for personalized and accessible breast imaging. The proposed project will improve the performance of a compact scanner and wearable accessory combination to produce repeatable images independent of operator training; this can be accomplished in under 2 minutes without expensive capital equipment, ionizing radiation, or patient discomfort. The intuitive software will enable physicians to visualize whole breast volume and accurately localize and measure lesions. AI will identify abnormal masses and predict the probability of malignancy to help physicians with accurate and fast diagnosis. The Phase II R&D focuses on five objectives: (i) optimize system performance for high-quality whole breast imaging with a new beamforming technique for higher resolution, with higher frame rates and faster scan; (ii) finalize the wearable and scanner design to ensure reliable operation with water as the coupling medium; (iii) conduct usability verification and validation regarding safety and functional requirements for clinical use; (iv) conduct a small study to verify the scanner's ability in finding existing breast lesions; (v) develop a machine learning engine for real-time detection and characterization of lesions in images acquired with the ultrasound scanner.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.