Deep Learning Applications in Ophthalmology

Ehsan Rahimy

Disclosures

Curr Opin Ophthalmol. 2018;29(3):254-260. 

In This Article

Limitations

Although there is a rapidly growing body of literature supporting a role for deep learning applications within ophthalmology, significant work remains as the next steps are taken towards its clinical validation and eventual implementation. Many of these studies utilized training sets from relatively homogenous patient populations. Moving forward, the goal will be to continue training on larger image sets, which are diverse across not only the patent demographic but also the type of images obtained (i.e. different fundus cameras, wide-field imaging, mydriatic versus nonmydriatic images, etc.). Ultimately, the algorithms learn from what they are presented with. Along these lines, efforts are being undertaken to help create more uniform reference standards amongst various graders and means for resolving grader disagreements, from which training of the algorithms occurs.[36] Furthermore, as may be expected, the algorithms appear to encounter difficulties whenever distinguishing potential artifacts from true disease that may be present (i.e. dust particles on a camera lens versus a potential microaneurysm/hemorrhage). Training these algorithms to infer when images are of substandard quality for grading is an area of ongoing research. Perhaps the greatest concern is the 'black box' nature of deep learning, whereby the rationale for the outputs generated by the algorithms are not entirely understood by not only the physicians but also the engineers who programmed them. This has created some apprehension in the public eye, and raises the potential dilemma of how to build public trust for something we do not fully comprehend. Nevertheless, groups have been attempting to fill in these gaps in knowledge by generating heat maps highlighting regions of influence on each image that contributed to the algorithm's conclusion.[22] Lastly, should we arrive at a future where automated image analysis has been integrated into clinical practice, there are concerns over whether this may eventually lead to a reduction in physician skills and clinical acumen because of an overreliance on technology.[37,38] This phenomenon is known as deskilling, where the skill level required to complete a task is reduced when components of the task become automated, leading to inefficiencies whenever the technology fails or breaks down.[37,38]

processing....