Rene Y. Choi, Aaron S. Coyner, Jayashree Kalpathy-Cramer
et al.
Purpose To present an overview of current machine learning methods and their use in medical research, focusing on select machine learning techniques, best practices, and deep learning. Methods A systematic literature search in PubMed was performed for articles pertinent to the topic of artificial intelligence methods used in medicine with an emphasis on ophthalmology. Results A review of machine learning and deep learning methodology for the audience without an extensive technical computer programming background. Conclusions Artificial intelligence has a promising future in medicine; however, many challenges remain. Translational Relevance The aim of this review article is to provide the nontechnical readers a layman's explanation of the machine learning methods being used in medicine today. The goal is to provide the reader a better understanding of the potential and challenges of artificial intelligence within the field of medicine.
Andy S. Huang, Kyle J. Hirabayashi, Laura E. Barna
et al.
Importance Large language models (LLMs) are revolutionizing medical diagnosis and treatment, offering unprecedented accuracy and ease surpassing conventional search engines. Their integration into medical assistance programs will become pivotal for ophthalmologists as an adjunct for practicing evidence-based medicine. Therefore, the diagnostic and treatment accuracy of LLM-generated responses compared with fellowship-trained ophthalmologists can help assess their accuracy and validate their potential utility in ophthalmic subspecialties. Objective To compare the diagnostic accuracy and comprehensiveness of responses from an LLM chatbot with those of fellowship-trained glaucoma and retina specialists on ophthalmological questions and real patient case management. Design, Setting, and Participants This comparative cross-sectional study recruited 15 participants aged 31 to 67 years, including 12 attending physicians and 3 senior trainees, from eye clinics affiliated with the Department of Ophthalmology at Icahn School of Medicine at Mount Sinai, New York, New York. Glaucoma and retina questions (10 of each type) were randomly selected from the American Academy of Ophthalmology's Commonly Asked Questions. Deidentified glaucoma and retinal cases (10 of each type) were randomly selected from ophthalmology patients seen at Icahn School of Medicine at Mount Sinai-affiliated clinics. The LLM used was GPT-4 (version dated May 12, 2023). Data were collected from June to August 2023. Main Outcomes and Measures Responses were assessed via a Likert scale for medical accuracy and completeness. Statistical analysis involved the Mann-Whitney U test and the Kruskal-Wallis test, followed by pairwise comparison. Results The combined question-case mean rank for accuracy was 506.2 for the LLM chatbot and 403.4 for glaucoma specialists (n = 831; Mann-Whitney U = 27976.5; P < .001), and the mean rank for completeness was 528.3 and 398.7, respectively (n = 828; Mann-Whitney U = 25218.5; P < .001). The mean rank for accuracy was 235.3 for the LLM chatbot and 216.1 for retina specialists (n = 440; Mann-Whitney U = 15518.0; P = .17), and the mean rank for completeness was 258.3 and 208.7, respectively (n = 439; Mann-Whitney U = 13123.5; P = .005). The Dunn test revealed a significant difference between all pairwise comparisons, except specialist vs trainee in rating chatbot completeness. The overall pairwise comparisons showed that both trainees and specialists rated the chatbot's accuracy and completeness more favorably than those of their specialist counterparts, with specialists noting a significant difference in the chatbot's accuracy (z = 3.23; P = .007) and completeness (z = 5.86; P < .001). Conclusions and Relevance This study accentuates the comparative proficiency of LLM chatbots in diagnostic accuracy and completeness compared with fellowship-trained ophthalmologists in various clinical scenarios. The LLM chatbot outperformed glaucoma specialists and matched retina specialists in diagnostic and treatment accuracy, substantiating its role as a promising diagnostic adjunct in ophthalmology.
This review is proposed to summarize the updates on COVID-19 and ophthalmology along with the bibliometric features of articles that have been published since the beginning of the COVID-19 outbreak. The databases, including PubMed, Scopus, and Web of Science, were searched using “Coronavirus,” “COVID-19,” “SARS-CoV-2,” “pandemic,” “ophthalmology,” “ophthalmic,” and “eye” keywords. All published articles except commentaries, errata, and corrigenda up to April 2021 were included. Titles and abstracts were screened, and ophthalmology-focused articles were collected. The bibliographic information of the articles, such as the name and country of the first author, type of study, date of publication, language, and journal name, were extracted. Included studies were assessed using the Joanna Briggs Institute (JBI) critical appraisal checklist. After systematic searching, 2,669 distinct articles were screened by title/abstract, and 1,174 ophthalmology-focused articles were selected to be reviewed. Ophthalmology-focused publications accounted for less than 0.5 percent of the total COVID-19-related articles. Most of the articles were published in the Indian Journal of Ophthalmology, and the main publication type was “original article.” Almost 88% of the publications were in English. There was a decline in the publication rate during the initial months of 2021 compared with the middle and last months of 2020. Most of the publications were affiliated with the United States of America. However, Singapore and the United Kingdom were the countries with the highest number of publications after population adjustment. Furthermore, a comprehensive review on major topics including SARS-CoV-2 ocular tropism, ophthalmic manifestations, ocular complications due to COVID-19 treatment strategies, the pandemic effect on ophthalmology care and operations, myopia progression during the pandemic, and telemedicine was conducted.
Kevin F. Elwood, Andrew C. Dieu, Clara V. Kuranz
et al.
Purtscher-like retinopathy (PUR) is a rare condition characterized by sudden vision loss with associated retinal white patches thought to be due to precapillary arteriolar occlusion. We present a case of PUR associated with a cardioembolic stroke in a patient following temporary cessation of anticoagulant therapy for a surgical procedure. Our patient presented with multiple risk factors for PUR and classic signs and symptoms including multiple peripapillary white retinal lesions near arterioles and sudden unilateral decrease in visual acuity. Optical coherence tomography showed inner retinal hyperreflectivity and thinning consistent with inner retinal ischemia, and fluorescein angiography showed delayed retinal filling. Her complement C5 factor was elevated on laboratory testing. Brain magnetic resonance imaging showed acute/subacute left occipital lobe ischemia thought to be from a cardioembolic stroke. Shortly prior to visual symptoms, our patient’s apixaban was held due to surgical drainage of a gluteal abscess. This case highlights the rare occurrence of PUR associated with cardioembolic stroke and the importance of cerebral imaging in a patient presenting with PUR of uncertain etiology.
Asher Khan,1,2 Kamran M Riaz,2 Neal Rangu,1,2 Vinay A Shah,2 Zain S Hussain,2,3 Mahmoud A Khaimi2 1College of Medicine, University of Oklahoma, Oklahoma City, OK, USA; 2Dean McGee Eye Institute, University of Oklahoma, Oklahoma City, OK, USA; 3University of Medicine and Health Sciences, Basseterre, Saint Kitts and NevisCorrespondence: Kamran M Riaz, Dean McGee Eye Institute, University of Oklahoma, 608 Stanton L. Young Blvd, Oklahoma City, OK, 73104, USA, Tel +1 405-271-1095, Fax +1 405-271-3680, Email Kamran-Riaz@dmei.orgIntroduction: Postoperative endophthalmitis (POE) is a rare but devastating complication of ophthalmic surgeries. Microinvasive glaucoma surgery (MIGS) procedures have become increasingly utilized for the surgical reduction of intraocular pressure (IOP). Ab-interno canaloplasty (ABiC) is a popular MIGS procedure, but POE rates and clinical effects following ABiC have not been studied.Methods: This study conducted a retrospective review of all consecutive cases of either standalone ABiC or combined ABiC with phacoemulsification performed at a tertiary care academic referral center from 2015 to 2021. Exclusion criteria included a history of incisional glaucoma surgery, retinal surgery, or additional concurrent microinvasive glaucoma surgery (MIGS) at the time of ABiC. The rates of POE after ABiC were calculated with 95% confidence intervals (CI) based on the Clopper-Pearson exact method.Results: Of 3256 cases of ABiC, one case (0.03%, 1/3256, 95% CI: 0.00– 0.17%) of post-ABiC endophthalmitis was identified. The rate of POE in standalone ABiC was 0.00% (0/1332 cases, 95% CI: 0.00– 0.28%), whereas the rate in combined ABiC with phacoemulsification was 0.05% (1/1924 cases, 95% CI: 0.00– 0.29%). Additionally, the rate of POE following stand-alone cataract surgery, 0.10%, 11/11,470 cases, 95% CI: 0.05– 0.17%), total cataract surgeries, 0.06% (17/28,013 cases, 95% CI: 0.04– 0.10%), total MIGs, excluding ABiC, surgeries, 0.08%, (3/3845 cases, 95% CI: 0.02– 0.23%) portray non-inferiority of ABiC in the risk of POE. The case of ABiC-POE presented four days after surgery and required a vitreous tap with intraocular injection of antibiotics and pars plana vitrectomy. No causative organism was identified. A final 1-year follow-up revealed a corrected distance visual acuity of 20/40 and stable glaucoma.Conclusion: The rate of POE after ABiC (1 per 3256 cases) is statistically non-inferior to the reported incidence of POE after other MIGS and incisional glaucoma surgeries.Keywords: ab-interno canaloplasty, canaloplasty, endophthalmitis, post-operative infections
AIM: To investigate the surgical outcomes of patients with chronic angle-closure glaucoma (CACG) treated with phacoemulsification (phaco)/endocyclophotocoagulation (ECP) with and without endoscopic goniosynechialysis (E-GSL). METHODS: A retrospective, nonrandomized, comparative case series was conducted. Patients with CACG who underwent phaco in combination with either ECP alone (ECP group) or GSL with ECP (E-GSL group) from 2018 to 2019 were followed for 12mo and reviewed. Clinical features and outcomes were identified and analyzed. The ECP and E-GSL groups were matched in age and baseline intraocular pressure (IOP). Changes in IOP, mean of visual acuity (VA), peripheral anterior synechiae (PAS) formation, and the number of glaucoma medications was examined. RESULTS: The ECP group included 32 eyes of 27 patients, and the E-GSL group included 32 eyes of 26 patients. The preoperative baseline IOP was 22.18±6.48 mm Hg in the ECP group and 22.95±6.71 mm Hg in the E-GSL group (P=0.644). The mean IOP reduction was 26.2% in the ECP group and 41.6% in the E-GSL group at 12mo. The mean postoperative VA (logMAR units) at 12mo was 0.47 in the ECP group and 0.36 in the E-GSL group. The reduction in PAS formation and the number of glaucoma medications was also higher in the ECP group than E-GSL group at 12mo. CONCLUSION: The phaco/ECP and phaco/E-GSL groups both achieve a significant reduction in IOP without complications associated with traditional glaucoma filtration surgeries.
Tanapat Ratanapakorn, Watcharaporn Thongmee, Kidakarn Meethongkam, Suthasinee Sinawat, Thuss Sanguansak, Chavakij Bhoomibunchoo, Wipada Laovirojjanakul, Yosanan Yospaiboon KKU Eye Center, Department of Ophthalmology, Faculty of Medicine, Khon Kaen University, Khon Kaen, ThailandCorrespondence: Yosanan YospaiboonKKU Eye Center, Department of Ophthalmology, Faculty of Medicine, Khon Kaen University, 123 Mitraparb Highway, Khon Kaen 40002, ThailandTel +66 43 348383Email yyosanan@gmail.comObjective: To study silicone oil (SO) emulsification, anatomic and visual outcome after complicated retinal detachment surgery by pars plana vitrectomy (PPV) with intraocular SO tamponade, comparing between low and high viscosity SO.Design: Randomized, double-blinded, controlled trial.Patients and Methods: Patients with complicated retinal detachment who had been surgically treated by PPV and intraocular SO tamponade were randomly divided into low viscosity (Group 1) or high viscosity (Group 2). Main Outcome Measures were rate of silicone oil emulsification, anatomic retinal reattachment and visual outcome, assessed at 1,3,6,9 and 12 months.Results: One hundred patients were divided into 50 patients in each group. The rate of silicone oil emulsification in Group 1 (63.64%) was higher than Group 2 (40%), but the difference was not statistically significant (p = 0.08). Rates of anatomic retinal reattachment were 88% in Group 1 and 90% in Group 2 (p = 0.76). For functional visual outcome, final best-corrected visual acuity (BCVA) at Month 12 compared with the baseline BCVA was improved vision 52% in Group 1 and 58% in Group 2 (p = 0.82).Conclusion: SO emulsification rate in low viscosity group is non-significantly higher than high viscosity group, but results in a comparable reattachment rate and final visual recovery. Emulsification in both groups is detected at as early as 1 month and mostly within 3 months. We recommend using either low or high viscosity SO in complicated retinal detachment surgery and removing it as early as possible to prevent the consequent serious complications.Clinical Trial Registration: ClinicalTrials.gov Identifier: NCT02988583.Keywords: silicone oil, viscosity, emulsification
Cumali Değirmenci, Melis Palamar, Nergis İsmayilova
et al.
Objectives:To compare data obtained by Scheimpflug camera (Pentacam) from both eyes of unilateral keratoconus patients and normal controls.Materials and Methods:This study was performed by retrospective chart review of 919 keratoconus patients. From these patients, 31 keratoconus eyes of 31 patients with unilateral keratoconus (Group 1), 31 normal fellow eyes of these patients (Group 2), and 30 right eyes of 30 normal controls (Group 3) were included in the study. Detailed ophthalmologic examination and Pentacam parameters at initial examination were analyzed and relationships between Groups 1, 2, and 3 were statistically evaluated. ROC curve analysis was also performed to determine the sensitivity and specificity of parameters that could be used to differentiate Group 2 from Groups 1 and 3.Results:The mean age was 30.07±11.00 (15-60) in Group 1-2 patients and 32.33±9.30 (18-45) in Group 3 patients (p=0.392). In comparison of Pentacam data, there were statistically significant differences between Groups 1 and 2 in all parameters except corneal volume (p<0.05). Group 1 and Group 3 were significantly different in all evaluated parameters (p<0.05). Steep keratometry, flat keratometry, mean keratometry, and posterior elevation (PE) were statistically similar between Groups 2 and 3 (p>0.05), while the other evaluated parameters differed significantly (p<0.05). ROC curve analysis showed that the difference in corneal thickness between the apex and thinnest point, progression index, index of surface variance, index of height asymmetry and inferior-superior had the highest sensitivity and specificity in differentiating Group 2 from Group 3, while CCTapex, CCTmin, PE, and minumum radius had the highest sensitivity and specificity in differentiating Group 2 from Group 1.Conclusion:In patients with unilateral keratoconus, fellow eyes appear to not be completely normal. Thus, it is recommended that fellow eyes also be evaluated in every examination of unilateral keratoconus patients.
A 25-year-old male patient presented at the clinic with chemical trauma to the left eye. His visual acuity according to Snellen chart measurement was 0.05/10 in the left eye and 10/10 in the right eye. On slit lamp examination, conjunctival hyperemia and limbal ischemia were observed between the 1 o'clock and 6 o'clock hours (approximately 5 clock hours). The cornea was edematous in this area, and a corneal epithelial defect adjacent to the limbal ischemia area was present. The anterior chamber, iris, and lens were normal; the fundus could not be concurrently evaluated. Conjunctivalization and excessive neovascularization were observed in the limbal ischemic area at the end of the fourth week of medical treatment. Visual acuity in the left eye was 0.3 and the epithelial defect had healed. At the end of the sixth week, a conjunctival limbal autograft (CLAU) combined with amniotic membrane (AM) transplantation surgery was performed. Subsequently, the corneal neovascularization was noted to have completely vanished in the ischemic area where the CLAU and AM transplantation were performed. Visual acuity improved to 0.7 and the symptoms of the patient decreased.
Stefan Nickels, Alexander K. Schuster, Heike Elflein
et al.
Abstract Purpose Most definitions of visual impairment focus on the status of the better-seeing eye only, but this approach might underestimate the influence of the worse-seeing eye on the vision-related quality of life (VRQoL). Methods We assessed distance-corrected visual acuity in both eyes and VRQoL using the “National Eye Institute 25-Item Visual Function Questionnaire” (NEI VFQ-25) in the German population-based Gutenberg Health Study. We calculated the Rasch-based visual functioning scale (VFS) and socioemotional scale (SES). We categorized the visual acuity of the better-seeing eye (BE) and worse-seeing eye (WE) as follows: (1) no visual impairment (VI) (< 0.32 logMAR)), (2) mild VI (0.32–0.5 logMAR), and (3) moderate to severe VI (> 0.5 logMAR). Next, the subjects were categorized as follows: both eyes with no VI (no/no), the better-seeing eye with no VI and the worse-seeing eye with mild VI (no/mild), no VI/severe VI (no/severe), both eyes with mild VI (mild/mild), light VI/severe VI (mild/severe), and both eyes with severe VI (severe/severe). We calculated the median scores for VFS and SES. We used linear regression to estimate the combined influence of BE/WE on VFS and SES. Results We included 11,941 participants (49.9% female, age range: 35–74 years) with information on VRQoL and visual acuity. The median VFS/SES scores were 90/100 (no/no VI group), 84/97 (no/mild group), 81/94 (no/severe group), 70/90 (mild/mild group), 67/74 (mild/severe group), and 63/76 (severe/severe group). These differences were supported by the regression analysis results. Conclusion Relying on the function of the better-seeing eye considerably underestimates the impact of visual impairment on VRQoL.
Computer applications to medicine. Medical informatics