The Scanner Room Gets a Brain
The MRI machine has been around since the 1970s. For most of that time, making a scan faster meant choosing between speed and image quality - the clinical equivalent of a coin flip. Enhao Gong arrived at Stanford to do his PhD and decided the coin flip was unnecessary.
His doctoral research under Professor John Pauly and neuroradiologist Dr. Greg Zaharchuk produced something the field didn't quite have yet: deep learning models trained specifically on medical imaging data, capable of denoising, accelerating, and enhancing MRI and PET scans without sacrificing the diagnostic detail clinicians depend on. The work was not a simulation. It was clinical-grade.
In late 2017, Gong and Zaharchuk co-founded Subtle Medical in Silicon Valley. The insight at the center of the company was elegant in its simplicity: every scanner in every hospital already generates enormous amounts of data. The bottleneck isn't hardware. It's the software sitting between raw signal and readable image. Gong built the software.
"If you only have a hammer of technology, and try to find the nail, it often won't work."- Enhao Gong, CEO, Subtle Medical
The Photo Editor Who Pivoted to Radiology
Before MRI acceleration, there was Polarr. In 2013, while deep in his Stanford PhD, Gong co-founded a photo editing startup alongside Borui Wang and Derek Yan. Polarr grew to tens of millions of users and eventually powered the camera software inside Samsung and Oppo phones. The technology worked. The market was real.
Gong chose to go back to finish his doctorate around 2015. That decision - returning to academic rigor while already proving commercial instincts - says something about how he operates. He is not in a hurry. He is precise.
The skills he developed building Polarr - training AI on image data, optimizing for visual quality, thinking about end-user experience - mapped directly to what Subtle Medical would require. The pivot wasn't random. The thread runs straight.
What 9 FDA Clearances Actually Look Like
Every product is vendor-neutral - meaning it runs on scanners from GE, Siemens, Philips, Canon, and others without requiring new hardware. Hospitals do not need to replace anything. They plug in the software. The scanner gets smarter.
Silicon Valley Speed Meets Regulatory Reality
Healthcare AI companies have a habit of moving fast and breaking regulations. Gong took a different path. From the start, he built Subtle Medical as a regulated medical device software company - one that would earn FDA clearances, not avoid them. The result is nine clearances that competitors cannot easily replicate.
In his Foothill Ventures interview, Gong described the tension at the core of healthcare startups: Silicon Valley wants you to ship, iterate, and break things. The FDA wants clinical evidence, safety validation, and a very thick file. He did not try to choose. He learned to navigate between the two.
"You have to know both, and you have to try to find a way to navigate between them."- Enhao Gong, on balancing startup velocity with healthcare regulation
This philosophy extends to co-founder selection. Zaharchuk brings 20+ years of clinical neuroradiology. Gong brings the AI and engineering. The company is not a technology company that happens to operate in healthcare. It was designed from day one to be bilingual: speaking fluent deep learning and fluent radiology simultaneously.
Forty-Five Papers and Counting
Academic publishing is not where most CEOs spend their time. Gong continues to do it. With 45+ peer-reviewed publications, his research sits at the intersection of computational imaging, deep learning, and clinical neuroscience - fields that rarely overlap in the same career, let alone the same paper.
Among the most-cited: a 2018 study in the Journal of Magnetic Resonance Imaging demonstrating that deep learning could reduce gadolinium contrast agent doses for brain MRI without compromising diagnostic quality. It was not theoretical. The results were reproducible. The FDA eventually cleared a product based on the principle.
Other research tackled quantitative susceptibility mapping using neural networks (QSMnet, published in NeuroImage 2018) and compressive sensing MRI via generative adversarial networks (IEEE Transactions on Medical Imaging, 2019). Each paper planted a flag in territory that Subtle Medical would eventually occupy commercially.
In 2023, Gong presented at the MICCAI Industrial Talk - a technical presentation on deep learning-based image enhancement, synthesis, and low-dose imaging. The YouTube recording remains a useful primer on what the company's AI actually does under the hood.
"Medical imaging is fundamental to clinical diagnosis. That's the technology we are developing to accelerate this process."- Enhao Gong
The Numbers That Matter Right Now
In 2025, Subtle Medical is not a startup story. It is a market story. The company reported 2.5x annualized revenue growth in the United States, earned CE Mark approvals to begin European expansion, and was named to TIME's World's Top HealthTech Companies list. Samsung Ventures backed the Series B, which exceeded $30 million.
The company's AI currently touches approximately 2.5 million patient scans per year across 1,000+ scanner installations globally. That number grows as each new hospital system signs on. The model scales without Gong flying to each hospital and installing hardware. Software at its most efficient.
Partnerships with NVIDIA accelerate the GenAI roadmap. An R01 grant supporting AI-powered pediatric imaging research with Boston Children's Hospital and Massachusetts General Hospital adds academic credibility alongside commercial traction. The two reinforce each other.
Gong's stated priority for the company's next phase: talent. "Finding talent is the most important aspect for us every day," he has said. With 70+ employees and sustained double-digit growth, the constraint is no longer technology or regulatory approval. It is people who can work at the speed his ambition requires.
From Tsinghua to Palo Alto via a Brain-Computer Interface on TV
Gong grew up watching a news segment about brain-computer interfaces that lodged in his memory and pointed his career. He studied biomedical engineering at Tsinghua University, one of China's most selective technical institutions, graduating in 2012. Then he moved to Stanford.
At Stanford, the doctoral work under Pauly and Zaharchuk gave him two things most AI founders lack: a deep understanding of imaging physics and a clinical co-investigator who would eventually become his company co-founder. The university incubated the relationship that the company runs on.
He is also affiliated with Stanford's Center for Artificial Intelligence in Medicine and Imaging (AIMI), which continues to surface his work to a research and clinical audience simultaneously.