We applaud Feng et al. on a well-conducted study that consolidates the advancements in the field of deep learning (DL) and its robust ability in the classification and interpretation of vitreous opacities in ophthalmic ultrasonography [1, 2]. This study evaluates several deep learning models like ResNet and GoogLeNet and their refined versions, identifying their capabilities in accurate image classification, thereby assisting in reducing misdiagnoses in ocular conditions and significantly improving diagnostic accuracies [1].

Terrestrial ocular and orbital ultrasonography (US) can detect abnormalities in the retina, vitreous, optic nerve, and orbit. The sensitivity and specificity of US however is technician and operator dependent and requires precision and accuracy as well as experience and expertise. Vitreous hemorrhages and retinal and choroidal detachment can be visualized on US when the view of the fundus is obscured [1]. US can also provide quantitative measurements of ocular and orbital tumors as well as the optic nerve and optic nerve sheath.

Log in or create a free account to read this content

Gain free access to this article, as well as selected content from this journal and more on nature.com

Access through your institution

or

Sign in or create an account
Continue with Google
Continue with ORCiD