Our Story

About the Founder

Krish Nachnani

Krish Nachnani is a student researcher focused on advancing health equity through AI and low-cost medical tools. He founded GlaucoScan.ai to bring early glaucoma screening to communities that lack access to specialists. His work combines machine learning, smartphone-based imaging, and clinical validation in real-world settings.

Why I Built GlaucoScan.ai

Living with progressive myopia and lattice degeneration, I’ve experienced the uncertainty of not knowing how my vision might change. That personal journey led me to study vision science and explore how technology could support earlier diagnosis for others.

While volunteering in Kenya, I saw how a lack of access to basic eye care often meant that glaucoma went undetected until vision loss had already occurred. In many communities, early screening simply wasn’t an option. I came back with a clear sense of urgency to build something practical that could help fill that gap.

A Low-Cost Screening Tool

GlaucoScan.ai is designed to meet that need. It combines low-cost hardware with AI models to make early glaucoma detection more accessible. The hardware uses a 3D-printed smartphone adapter, a Volk 20D lens, and the phone’s built-in camera and flashlight. This setup allows for clear retinal imaging without the need for bulky or expensive equipment.

Field Testing in India

To evaluate the tool in real-world conditions, I conducted pilot testing at two eye clinics in India: Ghaziabad Eye Hospital in Uttar Pradesh and Bhojay Sarvodaya Trust Clinic in Gujarat. More than 75 fundus images were captured and assessed for quality. These tests helped validate the system’s usability and confirmed that the hardware could reliably produce diagnostic-quality images in diverse clinical environments.

Screening a male patient using the 3D-printed adapter

Screening a female patient using the 3D-printed adapter

Improving the AI for Equity

The AI model behind GlaucoScan.ai has been continuously refined to ensure that it performs fairly across diverse populations. To improve equity across ethnicities, I modified the model using an approach called adversarial debiasing. This technique helps the AI focus on features relevant to glaucoma while minimizing the influence of demographic variables like race or ethnicity.

In addition to adversarial training, I’ve evaluated model performance across subgroups and iteratively updated datasets to support more balanced learning. This work is ongoing, with every version aimed at making the tool not only accurate but also equitable in how it serves patients from different backgrounds.

Key Publications

• Energy Efficient Learning Algorithms for Glaucoma Diagnosis — Published in IEEE Xplore
  https://doi.org/10.1109/ICMLA58977.2023.00307
• GAN-based Data Augmentation for Advanced Glaucoma Diagnostics — Featured in Recent Advances in Deep Learning Applications
  https://www.taylorfrancis.com/books/edit/10.1201/9781003570882

Presentations

Our work has been presented at leading scientific conferences, including:

• MIT Undergraduate Research Technology Conference
• IEEE International Conference on Machine Learning and Applications (ICMLA) 2023