I-Corps: Machine Learning-Based Diagnosis of Retinal Images with the Aim of Vision Loss Prevention

这个I-Corps项目的更广泛的影响/商业潜力是通过增加早期诊断检测的可及性来结束可预防的失明。利用新型成像技术和成熟的机器学习算法，该设备允许在初级保健环境中进行护理点诊断，成本不到当前成本的一半。通过使一线医生能够在不需要眼科护理专家的情况下提供挽救视力的眼科检查，拟议的技术可能会从根本上改变目前的视网膜检查格局，从而实现更有效和更容易获得的诊断测试。该I-Corps项目将产生一种基于人工智能（AI）的便携式视网膜成像系统，该系统由两个主要组件组成：（1）用于捕获患者视网膜图像的手持式眼科设备，以及（2）用于对视网膜图像进行分类的机器学习算法。硬件解决方案提供视网膜的高分辨率图像。红外照明对人眼是不可见的，并且消除了对瞳孔扩张的需要，瞳孔扩张在当前由医务人员进行的现有技术的眼底成像中被广泛使用。目前经过训练的卷积神经网络的准确率为97%，与经过训练的眼科医生的诊断准确率相当。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 to end preventable blindness by increasing access to early diagnostic testing. Utilizing a novel imaging technology and a proven machine learning algorithm, the device allows for point of care diagnosis in a primary care setting at less than half the current cost. By empowering frontline physicians to provide vision-saving eye exams without the need of an eye care specialist, the proposed technology may fundamentally change the current retinal exam landscape, resulting in more efficient and accessible diagnostic testing. This I-Corps Project will yield a portable Artificial Intelligence (AI)-based retinal imaging system consisting of two major components: (1) a hand held ophthalmic device to capture images of the patient's retina, and (2) a machine learning algorithm to classify images of the retina. The hardware solution provides high-resolution images of the retina. The infrared lighting is invisible to the human eye and eliminates the need for pupil dilation as is widely used in the state-of-art fundus imaging currently conducted by medical personnel. The current trained convolutional neural network yields an accuracy of 97% which is on par with the diagnostic accuracy of trained ophthalmologists.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.