To log in, click the teal "Login" button in the upper right-hand corner of this page. If you are logged in but still do not have access, please check your 2026 Annual Meeting registration.
Moderator
Thomas Kohnen, MD, PhD
Panelists
Anupama Janardhanan, MS, MBBS; Samina F Zamindar, MS, DOMS
Viewing Papers
Expand a paper title to the right to view the paper abstract and authors. Use the video link to jump to that poster in the session.
Presenting Author
Pradipta N. Chowdhury, MBBS
Co-Authors
Prakash Kumar Chowdhury (MS, DO), Anureet Kaur (MBBS)
Purpose
To develop and validate a clinical-grade AI model which can detect and stage cataracts from anterior selfies on smartphones. To enable autonomous cataract screening for early diagnosis without the need for slit-lamp imaging, attachments or clinical expertise.
Methods
This prospective study included 10,236 eyes clinically graded via slit-lamp as normal, immature and mature cataract. Anterior segment images were captured using front-facing smartphones under standardized lighting and preprocessed using CLAHE, HSV conversion, and red-channel extraction. YOLOv8 model was used to localize the pupil. A hybrid AI model combined Swin Transformer features with tabular metrics (brightness, red reflex percentage and entropy) for dual-output heads: categorical staging and continuous opacity scoring (0-5 scale). Adults ?18 with clear, clinically graded selfies were included; blurry/obscured pupils or ocular pathology without cataract were excluded.
Results
The model produced an overall accuracy of 96.0%, achieving 97.1% sensitivity (detecting cataract images) and 94.3% specificity (ruling out non-cataract images). The model's stage-wise classification displayed a strong correlation with clinical grading (r = 0.88, p < 0.001). The mean absolute error (MAE) for continuous opacity scoring was 0.21 with an R� of 0.85. The addition of tabular metrics (brightness, red reflex percentage and entropy) improved classification area under the curve (AUC) by 4.5%. Entropy showed the highest feature importance (SHAP value: 0.32). A high reliability score of 99.2% was achieved with YOLOv8 for pupil localization.
Conclusion
This model allows accurate detection and staging of cataract, simply from smartphone selfies - no slit-lamp, attachments, or training needed. It offers a low-cost, scalable solution for early diagnosis, improved outcomes and equitable vision care. The model enables screening from homes to even the most remote locations.
Presenting Author
Anupama Janardhanan, MS, MBBS
Purpose
To describe clinical characteristics, morphological types and visual outcome in pediatric posterior lenticular opacities operated at a tertiary care eye hospital in south India. To suggest a new classification of pediatric posterior polar cataracts
Methods
Retrospective study of pediatric posterior lenticular opacities that underwent cataract surgery over 5 years that presented to tertiary eye hospital were taken. Demographic details, Morphology, location of cataract, persistent hyaloid vasculature, anisometropia, amblyopia, strabismus and associated systemic conditions were assessed. One-year postoperative visual outcome was recorded.
Results
Total 121 pediatric posterior lenticular opacities met criteria.68% cases were unilateral with exotropia (17%). Visual outcome following cataract surgery was significant (p <0.01), improvement with amblyopia therapy was significant in bilateral cases (p=0.041). 93% intraocular lens was implanted in the bag. Anisometropia (18%) more common. Strabismus, PHPV, total cataract with ruptured lenticonus were associated with poor visual outcome.
Conclusion
We suggest a new classification of pediatric posterior lenticular opacities (SILC CLASSIFICATION) with 3 categories (SIMPLE, LENTICONAL, COMPLICATED POSTERIOR) and 9 subcategories for ease of diagnosis, surgical management, correlating systemic conditions and prognosis
Presenting Author
Vaishali Vasavada, MS
Co-Authors
Deepa Agarwal (MD), Shail Vasavada (DNB, FRCS), Abhay Vasavada (MS, FRCS), Samaresh Srivastava (MD)
Purpose
To investigate the long-term outcomes of congenital cataract surgery with IOL implantation atleast one decade after surgery
Methods
Prospective, interventional case series patients undergoing unilateral or bilateral congenital cataract surgery. Patients were followed up for atleast 10 years after the surgery. The outcome measures included major intraoperative complications, postoperative visual acuity, glaucoma, inflammatory complications and other postoperative complications following surgery in the long term.
Results
30 eyes of 21 patients were included in the study. The mean age at surgery was 47 months, with a mean follow-up duration 140 (standard deviation 11 months). Final results to be updated, ongoing study
Conclusion
To the best of our knowledge, this is the longest reported follow up of primary IOL implantation in Congenital cataract surgery. Based on preliminary analysis, it shows very good visual and technical outcomes with an acceptable rate of complications in these eyes
Presenting Author
Fatma F Shakarchi, MD, MSc
Co-Authors
Stephen LoBue (MD), Christopher Shelby (MD), Nicolas Zaunbrecher (MD), Curtis Martin (MD), Wyche Coleman (MD), Traeson Brandenburg (BSc), Ayorinde Cooley (MD, PhD)
Purpose
To highlight the role of open-access surgical courses in strengthening global ophthalmic education. We present a case study of a basic phacoemulsification course on Orbis Cybersight to support early surgical training for ophthalmologists worldwide.
Methods
We developed a one-hour, self-paced online course on phacoemulsification basics, divided into six narrated modules (Incisions, Capsulorhexis, Hydrodissection, Nucleus Removal, Cortex Removal, IOL Insertion & Wound Closure). Each module included quizzes for self-testing. The course was published on Orbis Cybersight, freely available on demand. We analyzed engagement patterns and collected informal feedback from learners and educators across regions.
Results
The course was accessed by >5,000 participants from 162 countries across all continents, including Antarctica, working in diverse high-, middle-, and low-income settings. Most were ophthalmologists (44%), residents (17%), and optometrists (10%), with others including medical students and nonmedical learners, including drivers, teachers, and security guards. About 27% completed the course, while 54% marked it "in progress." Average completion required ~84 minutes over ~5 attempts. Pre/post quizzes showed a 39.5% increase in participation and 35% relative knowledge gain.
Conclusion
Freely accessible platforms like Orbis Cybersight can deliver essential surgical training, which can bridge global gaps in surgical education, especially in areas with limitations in ophthalmologists' availability. Our case study offers a scalable, low-cost method to help global trainees gain exposure to essential surgical steps and concepts.
Presenting Author
Elad Ziv-On, MD, MPH
Purpose
The proliferation of artificial intelligence tools for health information seeking raises critical questions about their impact on patient adherence to medical recommendations. This study investigates whether AI tool usage affects patient compliance with post operative care instructions following cataract surgery, the most common ophthalmic surgery.
Methods
We are conducting a prospective observational study at Shaare Zedek Medical Center,Jerusalem, recruiting adults undergoing elective cataract surgery. At the one-month postoperative visit, participants complete a structured questionnaire assessing (1) AI tool usage patterns and health information-seeking behavior, and (2) adherence to postoperative instructions including topical medication regimens, activity restrictions, and follow-up visits. A composite adherence score (40% medications, 20% activity, 40% follow-up) serves as the primary endpoint. Multivariate analysis will examine associations between AI engagement and adherence, adjusting for demographic and clinical factors.
Results
TBD
Conclusion
TBD
Presenting Author
Soumya Gottipati, BSc
Co-Authors
Nakul Shekhawat (MD, MPH), Jordan Shuff (MSE), Kunal Parikh (PhD)
Purpose
Cataracts are the leading cause of blindness globally, but conventional slit lamps require trained personnel and are large, non-portable, and expensive, limiting accessibility in underserved settings. Here, we evaluate the usability and performance of a novel smartphone-based slit-lamp device, SeekerTM, for accessible cataract screening.
Methods
We developed a novel imaging device with a magnifying lens over a Samsung Galaxy S8 camera, a fixed slit-beam system (LED powered via smartphone OTG, a physical slit, and two cylindrical lenses) producing a 0.5 mm beam at 40°, and a scope to block ambient light. The device was piloted by a technician in a prospective study of dilated and non-dilated patients with pupil diameter ?4 mm, recruited from cornea and glaucoma clinics at the Wilmer Eye Institute. Images were graded by ophthalmologists for clinical diagnosability and six image quality criteria: focus, absence of external lighting artifacts, eye centration, slit beam pupil transection, primary gaze, adequate eyelid opening.
Results
Imaging was performed on 52 eyes (30 pseudophakia, 16 immature cataract, 6 clear lens) from 28 patients (median age 68.5; 57.1% female). Ophthalmologists rated 88.5% (46/52) of images as suitable for diagnosis. Image quality criteria were met as follows: focus (73.1%), no lighting artifacts (100%), centration (100%), slit transection (88.5%), primary gaze (100%), lid opening (94.2%). All six criteria were met in 65.4% of images. Cataract grading concordance between gold standard, in-clinic slit-lamp assessment and smartphone-based image review was 88.5%. All discordant cases were rated as clear lens via slit lamp examination, but were graded as immature cataract via SeekerTM image review.
Conclusion
The SeekerTM smartphone-based slit-lamp device enables acquisition of clinical grade, diagnostically useful anterior segment images suitable for telemedicine-based assessment of lens status, suggesting potential for deployment in underserved settings by non-specialists.
Presenting Author
Mohammadali Ashraf, MD
Co-Authors
Nicholas Tomaras (BSc), Hasan Khan (MD), Anubhav Pradeep (MS), Abdullah Ahmed (MBBS), Hajirah Saeed (MD, MPH), Mehmet Mocan (MD, MPH)
Purpose
Pediatric cataract is a leading, treatable cause of childhood blindness, where prompt diagnosis and coordinated care avert irreversible amblyopia. We aimed to define etiologies, demographics, and presenting features of dense pediatric cataracts and identify factors linked to severe visual impairment to guide care pathways.
Methods
We retrospectively studied all pediatric cataract encounters at a tertiary academic center from September 2017 to August 2024. Children with a clinical diagnosis of cataract were included and categorized by presenting visual acuity (case group: VA < 20/200 vs. control group: VA ? 20/200). Variables collected included demographics, insurance type, primary-care access, anthropometrics (height/weight percentiles), cataract etiology, laterality, coexisting ocular pathology (cornea, glaucoma, retina), and receipt of surgical management. The differences between groups were assessed and associations with dense cataract were estimated using logistic regression.
Results
Of 26,122 total cataract encounters, 189 (0.72%) were pediatric (47 cases, 142 controls). Mean age was 12.9 years in cases vs 11.5 in controls. Median weight/height were 40.5 kg/1.42 m in cases vs 28.1 kg/1.35 m in controls. Congenital etiology predominated in controls (77% vs. 57%) while trauma was higher in cases (21.3% vs 12.0%). Congenital etiology showed a protective trend against worse vision (OR 0.45). Though these were not statistically significant they are clinically significant differences. Cornea and glaucoma comorbidities were higher in cases, but only retinal comorbidity differed significantly (51% vs 33%; p=0.05) and was associated with worse vision (OR 2.04).
Conclusion
Trauma-related cataracts and retinal comorbidities were more common among children with dense cataracts, whereas congenital cataracts were linked to better visual outcomes. These findings highlight key risk factors for severe visual loss and can guide targeted care pathways to prevent lifelong vision impairment.
