Dr. Lamyanba Laishram
Talks about IBMS-Artificial Intelligence

Artificial Intelligence

Overview

Introduction

AI is transforming ophthalmology diagnostics and care. At IBMS, AI is applied across corneal and retinal imaging to surface findings that conventional methods miss and to accelerate clinical decision-making. In corneal work, the lab has used AI to identify vascular changes preceding tomographic changes in diabetic eyes, detect differential epithelial remodelling after LASIK, SMILE, and PRK, uncover unique corneal features of allergic eye disease, and combine corneal confocal microscopy with clinical features to evaluate ocular surface pain.

Why This Research Matters

AI has also identified regional differences in tomographic parameters of keratoconic corneas and powered a universal architecture for early keratoconus diagnosis. The lab’s most recent work uses AI-based stratification of high-risk demographic and ocular-surface factors to study keratoconus progression. In retinal work, the lab builds intelligent systems that assist clinicians by analysing retinal images and highlighting important diagnostic patterns, with a focus on expanding access to specialist-level screening.

Objectives

The Artificial Intelligence programme aims to:

  • Surface findings that conventional methods miss across corneal and retinal imaging
  • Accelerate clinical decision-making through AI applications
  • Identify vascular changes preceding tomographic changes in diabetic eyes
  • Power universal architecture for early keratoconus diagnosis
  • Develop intelligent systems for retinal image analysis and specialist-level screening

Core Research Focus

The artificial intelligence programme focuses on understanding AI applications in ophthalmology diagnostics and care.

Core areas of focus include:

  • Diabetic Eyes: AI detection of vascular changes that precede visible tomographic changes
  • Refractive Surgery: Differential epithelial remodelling patterns after LASIK, SMILE, and PRK identified by AI
  • Allergic Eye Disease: AI identification of unique corneal features not visible through standard examination
  • Ocular Surface Pain: Novel diagnostic approach combining corneal confocal microscopy, clinical features, and AI
  • Keratoconus: Regional tomographic pattern analysis, universal early diagnosis architecture, and demographic and ocular surface risk stratification for progression monitoring
  • Retinal Disease: Multimodal LLM system for automated interpretation and clinical report generation from ultra-widefield retinal images, trained on real-world Indian patient data and designed for cloud deployment

Key technical directions include:

  • Early detection of vascular changes in diabetic eyes preceding tomographic alterations
  • Differential epithelial remodelling identification post-refractive surgery (LASIK, SMILE, PRK)
  • Unique corneal features of allergic eye disease uncovered by AI.
  • Ocular surface pain evaluation combining confocal microscopy and clinical features.
  • Regional tomographic differences and progression risk stratification in keratoconus.

Together, these efforts support a diverse portfolio of Artificial Intelligence research, with support from Engineers, Clinicians, and Imaging Scientists.

Team

The Artificial Intelligence programme is led by a multidisciplinary team of scientists and clinician researchers with expertise in:

  • Engineers
  • Clinicians
  • Imaging Scientists

Close collaboration between laboratory researchers and clinicians ensures that research priorities remain aligned with patient needs and real-world clinical applicability.

NVAULT - Artificial Intelligence

Non-contact quantification of topography of anterior corneal surface and Bowman's layer with high speed OCT

Non-contact quantification of topography of anterior corneal surface and Bowman’s layer with high speed OCT

Authors: Matalia H, Francis M, Gangil T, Chandapura R, Kurian M, Shetty R, Nelson E, Sinha Roy A

Keywords: NA

PMID: 28486724

DOI: 10.3928/1081597X-20170201-01

Quantitative evaluation of OCT angiography images of diabetic retinopathy eyes before and after removal of projection artifacts.

Quantitative Comparison of Retinal Vascular Features in Optical Coherence Tomography Angiography Images from Three Different Devices

Authors: Anegondi N, Kshirsagar A, Mochi TB, Sinha Roy A

Keywords: NA

PMID: 30021035

DOI: 10.3928/23258160-20180628-04

Outcomes of Corneal Cross-Linking Correlate with Cone-Specific Lysyl Oxidase Expression in Patients with Keratoconus.

Outcomes of Corneal Cross-Linking Correlate with Cone-Specific Lysyl Oxidase Expression in Patients with Keratoconus.

Authors: Shetty R, Rajiv Kumar N, Pahuja N, Deshmukh R, Vunnava K, Abilash VG, Sinha Roy A, Ghosh A.

Keywords: NA

PMID: 29215396

DOI: 10.1097/ICO.0000000000001478

See All

Future Vision

The AI and Retinal Imaging programme is progressing towards:

  • Advancing AI-driven early diagnosis of keratoconus, risk modelling for progression, and detection of allergic eye disease.
  • Developing multimodal large language model (MLLM) systems for automated retinal image understanding and clinical report generation.
  • Deploying ultra-widefield imaging models (capturing up to 200° of the retinal surface) that produce structured clinical descriptions covering disease presence, anatomical location, severity, and diagnostic context.
  • Enabling cloud-based deployment to support large-scale screening where specialist access is limited
  • Building AI systems that work alongside clinicians to reduce turnaround time, expand access to specialist-level screening, and improve patient outcomes where expert capacity is limited.