I-Corps: Automatic Aortic Aneurysm Screening using Deep-Learning Models

该I-Corps项目更广泛的影响/商业潜力是为放射科医生和血管外科医生开发软件平台，以帮助减少筛查主动脉瘤所需的时间。 所提出的技术旨在提高计算机断层扫描（CT）扫描读数、三维（3D）图像重建和主动脉瘤筛查的速度和效率。 目前，注释和测量主动脉瘤是手动完成的，并且是耗时的。 该技术的自动筛查功能可以提高医生的工作效率，并改善患者体验。 除了主动脉瘤之外，自动筛选特征可以用于医学筛选领域中的其他应用，包括脑肿瘤、肺栓塞和肺癌。该项目的成果和资源也可用于教育目的，新的课程将根据该项目开发，涵盖神经网络应用，三维（3D）图像重建，这个I-Corps项目的基础是开发一个软件平台，该平台可以自动读取计算机断层扫描（CT）扫描使用算法来训练深度学习模型，并允许有效和准确地测量主动脉瘤的直径。特别是，所提出的技术使用基于亮度和对比度的自适应图像处理方法来微调深度学习模型，并采用基于集成的最先进的客观检测模型。所提出的技术可以呈现主动脉的3D可视化，这可以帮助医生对动脉瘤进行更全面的诊断。此外，基于测量值和患者健康病历数据，所提出的方法被设计用于计算主动脉不良事件的估计风险该奖项反映了NSF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 impact/commercial potential of this I-Corps project is the development of a software platform for radiologists and vascular surgeons to help reduce the time required for screening aortic aneurysms. The proposed technology is designed to improve the speed and efficiency in computed tomography (CT) scan readings, three-dimensional (3D) image reconstruction, and aortic aneurysm screening. Currently, annotating and measuring an aortic aneurysm is done manually and is time consuming. The proposed technology’s automatic screening feature may increase the productivity of doctors who routinely read the scans and improve patient experiences. In addition to aortic aneurysms, the automatic screening feature may be used in other applications in the medical screening field including brain tumors, pulmonary embolisms, and lung cancer. The results and resources developed from this project also may be used for educational purposes, where new curricula will be developed based on this project to cover the fields of neural network applications, three-dimensional (3D) image reconstruction, and image processing algorithms.This I-Corps project is based on the development of a software platform that automates the readings of computed tomography (CT) scans using algorithms to train deep-learning models and allow for efficient and accurate diameter measurement of aortic aneurysms. In particular, the proposed technology uses adaptive image processing methods based on brightness and contrast to fine tune deep-learning models and employ ensemble-based state-of-the-art objective detection models. The proposed technology may render 3D visualizations of the aorta, which may help doctors to perform more comprehensive diagnosis of the aneurysm. Moreover, based on the measurements and patient health medical record data, the proposed method is designed to calculate estimated risk of adverse aortic events (dissection, rupture, and death), and report risk-rankings to help prioritize patient treatment and surgical procedures.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.