Inventor recalls eye imaging breakthrough
David Huang, an MIT-trained clinician-scientist, invented optical coherence tomography in 1991, and the technology now powers 40 million eye imaging procedures every year. Researchers are now layering artificial intelligence on top of OCT to make retinal scans 100 times faster an...
According to MIT Technology Review, David Huang, who earned his undergraduate degree from MIT in 1985, his master's in 1989, and his PhD in 1993, recently reflected on how a graduate school experiment turned into one of the most widely used medical imaging tools on the planet. The piece traces OCT from its origins in a laser research lab to its current role in ophthalmology offices worldwide, and it arrives at a moment when AI researchers are actively rebuilding what the technology can .
Why This Matters
OCT is not a niche research curiosity. It runs in 40 million procedures annually, which means it has touched more patients than most medical technologies invented in the last half-century. The fact that AI labs at the National Institutes of Health and UC Berkeley are now treating OCT as a foundation to build on, rather than a finished product to replace, tells you everything about how durable Huang's original insight was. When a 1991 invention becomes the substrate for 2025 AI research, that is not nostalgia. That is a platform.
Daily briefing from 50+ sources. Free, 5-minute read.
The Full Story
David Huang did not walk into MIT planning to reinvent eye imaging. He enrolled as an electrical engineering undergraduate with a simpler motivation: he wanted to apply engineering thinking to medicine, the way his father had applied medical thinking to patient care as a family practitioner. That instinct to bridge disciplines would define everything that followed.
The breakthrough came during Huang's time as an MD-PhD student in the Harvard-MIT Program in Health Sciences and Technology. He was working in the lab of James Fujimoto, the Elihu Thomson Professor of Electrical Engineering at MIT, who had tasked him with using ultrafast lasers to improve ophthalmological measurements, specifically the thickness of the cornea and retina. Huang identified interferometry as the right tool for the job. Interferometry can measure the time of flight of light waves down to one quadrillionth of a second, which translates to micrometer-level resolution in tissue imaging.
What Huang discovered in his experiments was that interferometry could detect extremely faint signals from fine internal structures inside the retina, structures that existing imaging methods could not resolve clearly. Fujimoto recognized immediately that they were looking at the foundation of a new imaging modality. The two brought in Eric Swanson, who had earned his master's at MIT in 1984 and was at the time using interferometry for intersatellite communications research at Lincoln Laboratory, to help build an actual OCT machine for biological use.
Huang then tested the device on tissue samples obtained through Harvard Medical School. The results were strongest with retinal tissue and coronary artery samples, two areas where high-resolution internal imaging had enormous clinical value. The team published their findings in Science in 1991, formally establishing OCT as a new category of medical imaging. What separated their work from others exploring similar territory, Huang has said, was access: MIT's advanced laser technology, Lincoln Lab's interferometry expertise, and Harvard's medical connections created a combination that competing researchers could not match.
After the Science publication, Huang completed his ophthalmology training while Fujimoto and Swanson launched a startup to commercialize the device and get it into clinical settings. The transition from lab invention to standard clinical tool took years, but OCT eventually became the default imaging method for conditions including macular degeneration, diabetic retinopathy, and glaucoma. In 2023, Huang and his co-inventors received the Lasker Award and the National Medals of Technology and Innovation. In 2025, all three were inducted into the National Inventors Hall of Fame.
Key Details
- OCT is now used in 40 million procedures per year globally.
- David Huang earned his BS from MIT in 1985, his SM in 1989, and his PhD in 1993.
- The original OCT paper was published in Science in 1991.
- Co-inventors include James Fujimoto (MIT class of 1979, SM 1981, PhD 1984) and Eric Swanson (SM 1984).
- Huang, Fujimoto, and Swanson received the Lasker Award and National Medals of Technology and Innovation in 2023.
- All three inventors were inducted into the National Inventors Hall of Fame in 2025.
- OCT uses infrared light and produces three-dimensional images at higher resolution than traditional fundus photography.
What's Next
The next phase for OCT is less about the hardware and more about the software layered on top of it. NIH researchers have demonstrated that AI can accelerate retinal image acquisition and analysis by roughly 100 times compared to manual methods, and that same AI pipeline improves image contrast by 3.5 times. Watch for clinical trials that attempt to validate retinal AI screening as a tool for detecting early cognitive decline, a research direction that could expand OCT's use far beyond ophthalmology departments.
How This Compares
The NIH's work using a technique called P-GAN to speed up retinal imaging by 100 times is the most direct extension of what Huang built. It does not replace OCT. It makes OCT dramatically more efficient, which is exactly how transformative platform technologies tend to evolve. Compare this to how MRI hardware developed over decades before software and AI made interpretation faster and more consistent. OCT appears to be following the same arc, just on a compressed timeline because of how rapidly AI tooling has matured. You can follow related AI tools making similar leaps in medical imaging on this site.
Research published in September 2025 showed that AI-powered eye scans can predict the risk of cognitive decline and dementia, which is a genuinely unexpected extension of a technology originally designed to measure retinal thickness. That finding, combined with collaborative work at UC Berkeley and UCSF applying AI to medical imaging more broadly, suggests the field is converging on a vision of the eye as a diagnostic window into systemic health. That is a significant expansion of scope from what Huang published in 1991.
Where this diverges from, say, the wave of AI diagnostic startups pitching entirely new imaging hardware is instructive. Huang's original technology already exists in virtually every ophthalmology practice in the developed world. AI researchers are not asking clinicians to buy new equipment. They are making the equipment clinicians already own dramatically more powerful. That is a far easier adoption path than asking hospitals to replace capital equipment, and it is why OCT-plus-AI deserves more attention than it currently gets in AI coverage circles. For broader context on where medical AI is heading, the AI Agents Daily news section tracks these developments regularly.
FAQ
Q: What is optical coherence tomography and how does it work? A: OCT is a noninvasive imaging technique that uses infrared light to produce detailed, three-dimensional images of internal tissue structures like the retina. It measures the time it takes for light waves to reflect off tissue at a precision of one quadrillionth of a second, which allows it to create high-resolution cross-sections without touching the eye or using radiation.
Q: Who invented OCT and when was it created? A: David Huang, working with James Fujimoto and Eric Swanson at MIT and Lincoln Laboratory, invented OCT as part of his MD-PhD research. They published their initial findings in the journal Science in 1991. Huang and his co-inventors received the Lasker Award and National Medals of Technology and Innovation in 2023 and were inducted into the National Inventors Hall of Fame in 2025.
Q: How is AI being used to improve eye imaging today? A: NIH researchers have applied AI to retinal imaging using a technique called P-GAN that accelerates image acquisition by approximately 100 times compared to manual methods and improves image contrast by 3.5 times. Separate research published in September 2025 showed that AI-analyzed eye scans may also predict the risk of cognitive decline, opening up entirely new clinical applications beyond ophthalmology.
David Huang's story is a reminder that the most consequential inventions are rarely the ones that arrive fully formed. OCT spent decades becoming indispensable before AI researchers decided it was worth supercharging. The next 10 years of retinal imaging will likely look nothing like the last 30, and the foundation Huang laid in 1991 will be the reason why. Subscribe to the AI Agents Daily weekly newsletter for daily updates on AI agents, tools, and automation.
Get stories like this daily
Free briefing. Curated from 50+ sources. 5-minute read every morning.
