[Artificial intelligence in neuro-ophthalmology: prospects, challenges and countermeasures].

Neuro-ophthalmic disorders feature complex etiology. Certain ocular manifestations may hint at severe neurological diseases. Current clinical practice faces no shortage of diagnostic examinations. The core challenge consists in inadequate recognition, integration and interpretation of multifaceted information during initial consultation and non-specialist visits, potentially resulting in misdiagnosis, missed diagnosis, delayed referral and excessive examinations. Artificial intelligence has demonstrated promising performance in optic disc lesion detection, image interpretation of optic neuropathy, visual field analysis, as well as eye movement and pupillary function evaluation. Nevertheless, AI is not intended to replace specialists to deliver definitive diagnoses. Instead, it helps boost early detection of high-risk signs, facilitates analysis of complicated test results, and aids clinical decision-making, patient referral and follow-up management. This article elaborates on the application prospects, existing dilemmas and translational strategies of artificial intelligence in neuro-ophthalmology, aiming to offer references for relevant research and clinical practice.