Publications
Alongside our product development efforts, we also support research that turns the burden of corneal disease in low- and middle-income countries into evidence, tools, and better outcomes. The publications, pre-prints, and conference proceedings below were made possible through KeraLink International's funding of KeraX projects and partners.
Applications of Computer Vision for Infectious Keratitis: A Systematic Review
Assaf JF, Ahuja AS, Kannan V, Yazbeck H, Krivit J, Redd TK.
Ophthalmology Science. 2025;5(6):100861.
KEY FINDINGS
A systematic review of 37 studies (2017–2024) applying artificial intelligence and computer vision to infectious keratitis, spanning bacterial–fungal differentiation and pathogen detection. It maps the state of the field, highlights gaps in dataset diversity and external validation, and sets priorities for building AI that works across the low- and middle-income settings where corneal blindness is concentrated.
ROLE
Co-Author
Anatomic Biomarkers Predict Poor Presenting Visual Acuity in Infectious Keratitis
Anant S, Reddy K, Shuff J, Parikh KS, Jain E, Kuyyadiyil S, Parmar G, Shekhawat NS.
Clinical Ophthalmology. 2025;17:4155–4168.
KEY FINDINGS
A cross-sectional study of adults with active infectious keratitis at Sadguru Netra Chikitsalaya (SNC) Hospital in rural Madhya Pradesh, India, examined which findings at presentation predict severe vision loss. Anatomic signs visible on routine slit-lamp examination — hypopyon among them — were associated with worse presenting visual acuity, giving front-line clinicians a practical way to stratify keratitis severity where advanced diagnostics are unavailable.
ROLE
Funder
Detection and Measurement of Hypopyon on Slit Lamp Examination Versus Anterior Segment Optical Coherence Tomography
Reddy KN, Ibukun F, Huang K, Yi J, Jain E, Kuyyadiyil S, Parmar G, Shekhawat NS.
Bioengineering. 2026;13(5):582.
KEY FINDINGS
Hypopyon — a layer of inflammatory cells in the anterior chamber — is an important marker of keratitis severity. This study compared anterior-segment OCT (AS-OCT) with standard slit-lamp examination and found AS-OCT detected hypopyon in more eyes (67% vs 57%) and produced highly reproducible measurements between graders, supporting AS-OCT as a sensitive, objective tool for grading anterior-chamber inflammation.
ROLE
Funder
Deep Learning for the Detection of Corneal Perforation on Anterior-Segment Optical Coherence Tomography in Microbial Keratitis
Rhode LH, Reddy KN, Ibukun F, Kuyyadiyil S, Jain E, Parmar GS, Chellappa R, Shekhawat NS.
Bioengineering. 2026;13(6):649.
KEY FINDINGS
Corneal perforation is a sight-threatening emergency in microbial keratitis. Using AS-OCT scans from 150 patients, the team trained deep-learning models (a ResNet-34 backbone over six radial scans per eye) to automatically flag perforation. AS-OCT identified perforation in roughly twice as many eyes as slit-lamp examination, pointing toward scalable, expert-level triage for settings where cornea specialists are scarce.
ROLE
Funder
Detection of Infectious Corneal Perforation Using Anterior Segment Optical Coherence Tomography
Ibukun F, Reddy K, Kuyyadiyil S, Jain E, Parmar G, Shekhawat NS.
medRxiv (preprint). 2026
KEY FINDINGS
A diagnostic-accuracy study of 150 eyes evaluating whether AS-OCT detects corneal perforation more reliably than slit-lamp examination, and how reproducible AS-OCT grading is between masked experts. AS-OCT served as a sensitive adjunct that caught perforations missed on clinical examination — with implications both for patient care and for the reliability of perforation as an endpoint in corneal clinical trials.
ROLE
Funder
Rethinking Multiple Instance Learning for Corneal Perforation Detection on Radial Anterior Segment Optical Coherence Tomography in Microbial Keratitis
Rhode LH, et al.
Medical Imaging with Deep Learning (MIDL) 2026 — Short Paper Track
KEY FINDINGS
A machine-learning methods study on how best to combine the multiple radial AS-OCT cross-sections captured per eye into one reliable perforation prediction. It reframes corneal-perforation detection as a multiple-instance learning problem, informing the design of robust, data-efficient AI for anterior-segment imaging.
ROLE
Funder
Demographic, Clinical, and Anatomic Risk Factors for Keratitis Severity in an Indian Population
Anant S, Reddy K, Shuff J, Parikh KS, Jain E, Kuyyadiyil S, Parmar G, Shekhawat NS.
ARVO Annual Meeting Abstract, June 2025. Investigative Ophthalmology & Visual Science.
KEY FINDINGS
Presented at the ARVO 2025 Annual Meeting, this analysis of the SNC Hospital cohort examined how demographic, clinical, and anatomic factors relate to keratitis severity at presentation in a rural Indian population — the groundwork for the peer-reviewed findings on anatomic biomarkers of poor visual acuity above.
ROLE
Funder