Neuro-Ophthalmic Literature Review

Abstract

Neuro-Ophthalmic Literature Review David A. Bellows, Noel C.Y. Chan, John J. Chen , Hui-Chen Cheng, Jenny A. Nij Bijvank, and Michael S. Vaphiades New Findings Contradict Previous Credenda Regarding Paediatric Optic Neuritis Pineles SL, Repka MX, Liu GT, Waldman AT, Borchert MS, Khanna S, Heidary G, Graves JS, Shah VS, Kupersmith MJ, Kraker RT. Assessment of paediatric optic neuritis visual acuity outcomes at 6 months. JAMA Ophthalmol. 2020;138(12):1253– 1261. doi:10.1001/jamaophthalmol.2020.4231 This prospective study was the fruit of the joint efforts of the Paediatric Eye Disease Investigator Group and the Neuro-Ophthalmology Research Disease Investigator Consortium at 23 academic and community-based clinical sites. Forty-four children aged 3 to 15 years were entered into the study of which 37 completed the six-month follow-up. Sixteen (36%) of the children presented with bilateral optic neuritis and the remainder were unilateral. Optic disk oedema was seen in 41 eyes (75%) and retinal haemorrhages in two eyes (4%). Magnetic resonance imaging (MRI) revealed optic nerve enhancement in 36 (92%) children and white matter lesions in 23 (52%). Fifteen of the children were tested for NMO antibodies and, of these, only one (7%) tested positive. Thirteen of the enrollees were tested for MOG antibodies and, in this group, seven (54%) tested positive. The mean distance visual acuity of 54 eyes at enrolment was 20/200. Twenty-eight eyes had a visual acuity of less than 20/200 and 16 eyes were less than 20/800. Factors associated with poor visual acuity included younger age, nonWhite and non-Hispanic ethnicity, an associated neurological autoimmune diagnosis (such as ADEM, NMO or MOG), and the presence of brain lesions on MRI. Thirty-seven patients remained in the study at 6 months. Their mean improvement in visual acuity was eight lines on a standard ETDRS chart. Thirtyfour (77%) of the children fell within their agenormal range and two children (4%) had a visual acuity worse than 20/200. These data contradict the standard teachings that most cases of paediatric optic neuritis are bilateral and neurologically isolated. David A. Bellows More AI coming up: Differentiating NGON from GON Yang HK, Kim YJ, Sung JY, Kim DH, Kim KG, Hwang J-M. Efficacy for differentiating nonglaucomatous versus glaucomatous optic neuropathy using deep learning systems. Am J Ophthalmol 2020; 216:140–146. doi:10.1016/j.ajo.2020.03.035 The investigators of this single institute developed an Artificial Intelligence Classification algorithm to differentiate images with normal optic discs, glaucomatous optic neuropathies (GON) and non-glaucomatous optic neuropathies (NGON). Among 3,815 fundus images collected, there were 486 GON images and 446 NGON images where the rest were normal optic disc images. Diagnosis of NGON included compressive optic neuropathy, Leber hereditary optic neuropathy, autosomal dominant optic atrophy, toxic and traumatic optic neuropathy, as well as optic atrophy of unknown cause. They reported a high overall diagnostic accuracy of 99.1%. The accuracies of detecting normal discs, NGON, and GON were 99.7%, 86.4%, and 92.5%, respectively. The sensitivity and specificity in differentiating GON from NGON images were 92.5% and 99.5%, respectively, with an average precision of 0.954. The major reasons for false positive findings were peripapillary atrophy and tilted optic discs. CONTACT John J. Chen [email protected] Department of Ophthalmology, Mayo Clinic, 200 First Street, SW, Rochester, MN 55905, USA. NEURO-OPHTHALMOLOGY 2021, VOL. 45, NO. 1, 68–73 https://doi.org/10.1080/01658107.2021.1877990 © 2021 Taylor & Francis Group, LLC Limitations of this study include the recruitment of variable stages of GON and NGON. There was also a significant portion (n = 73) with unknown cause of optic atrophy in the NGON group. Nevertheless, the study has certain strengths. High intraocular pressure was not a criterion in the inclusion of GON images; thus, optic discs of normal tension glaucoma (NTG) were also evaluated though the proportion of which was not reported. In real-life practice, it is still controversial to obtain neuroimaging for all NTG patients and differentiating NGON from NTG may pose diagnostic challenges to many. Secondly, the deep learning model also showed good performance without prior control of image acquisition parameters. This makes the system capable of a broader application on images with variable qualities in the future. This study demonstrates great potential in employing artificial intelligence in the future screening programme. This technology is also helpful in assisting general ophthalmologists to differentiate NGON from GON to decide whether urgent neuroimaging or investigations are required. However, further validation of the system in different populations and ethnicities is necessary. This is particularly important as the colour of the fundus photography may depend on the degree of choroidal pigmentation. Lower accuracy is also expected in regions with a high population of high myopia.