Special Edition: Why we invested in Hypervision Surgical. The future of AI-driven surgery.
Hypervision Surgical just announced their GBP 17m Series A. Let's talk about what they're building, why we love it and why it's essential for the future of automated surgery.
Surprise! This article hits your inbox early (today instead of Monday). The reason: Yesterday, we announced our investment in Hypervision Surgical.
They raised a GBP 17m Series A led by us (Heal Capital) and joined by many great funds and partners. Click here for the founder’s official announcement, including all those brilliant people.
We’re super hyped - and we think you deserve to know more.
This will be a new series with the straightforward name “why we invested”. You get the idea: We share what convinced us to invest in the specific team, market and timing. We will write these pieces for all upcoming investments, so stay tuned. It’s one of the scariest things VCs can do: put their believes out in public, create an audit trail of their decisions and be held accountable for them later.
I promise we’ll deliver:
an explanation of what the company actually does, incl. the vision, in plain words
facts you won’t find in the official press releases
no PR-washed impact bingo or namedropping
Ready? Let’s dive into Hypervision Surgical.
From here on, I’ll give the stage to my colleague Felix who is closest to the deal. He’ll explain in his words:
A glimpse at the status quo
Walk into an operating room during a bowel cancer procedure and you’ll see something that hasn’t really changed in decades. The surgeon removes the tumor, then has to rejoin the two cut ends of intestine. Whether that join heals depends on many things, but one of the most essential is whether the tissue at the edges is still getting enough blood. The way the surgeon decides where to cut is by looking. They watch the colour, the sheen, how a small vessel bleeds when nicked. They’ve seen thousands of these tissues before, and over time that pattern recognition becomes something close to intuition.
It is genuinely impressive to watch. Unfortunately, in about one out of every ten cases, things go wrong. The join breaks down, the patient gets an infection, ends up back in surgery, sometimes in the ICU, sometimes worse. Decisions with life-changing consequences for the patient are still made by eye.
Surgery is unusual in this. Almost every other corner of medicine has gradually moved from judgment to measurement. Cardiologists used to read short ECG strips, now they read days of continuous data. Oncologists used to pick treatment by where the tumor sat, now they sequence the cancer’s genome. Radiologists work alongside software that quantifies what they see. Surgery, where the stakes are incredibly high for the patient, has stayed largely qualitative.
The limits of looking
There is a partial fix already in use. A surgeon can inject a fluorescent dye into the bloodstream, switch the camera to near-infrared mode, and watch the well-supplied tissue glow. The evidence behind it is good and it has become the accepted way to check blood flow during surgery. In practice it is more limited than it sounds. Each injection gives one brief look. A surgeon can only do this a few times in a case. And even then, the final call still comes down to a person making a visual judgement about brightness. Worldwide, this technique gets used in fewer than one in ten procedures where it could help. The rest of the time, it is the surgeon’s eyes again.
You might assume this is a software problem waiting on better AI. The reason this has been so hard in practice has less to do with the model and more to do with what the camera gives the model to work with.
Surgical cameras, like nearly every camera ever built, only capture the red, green, and blue colour information the human eye can see. They were designed to show a person what is in front of the lens. Anything outside those three colours is filtered out by the sensor before any software gets a chance at it. Researchers have made real progress trying to estimate things like tissue oxygenation from RGB alone, but getting enough validated training data, and proving the models perform reliably inside an operating room, has so far been prohibitive.
The physics for actually measuring it directly have been settled for half a century. Oxygenated and deoxygenated blood absorb light differently at specific wavelengths, but the difference is far too subtle for the naked eye to read reliably. The pulse oximeter on a hospital patient’s fingertip uses the same principle to read oxygen levels in the blood. What has never existed in surgery is a way to apply that physics across an entire field of view, on tissue that is moving, while the surgeon keeps working and cannot wait.
Seeing what the eye can’t
Hypervision Surgical introduced a new way to sense tissue in surgery, one that goes beyond what the human eye can take in. Their camera captures dozens of narrow wavelength bands across the spectrum of light instead of just three. Every pixel carries a small spectrum of information about the tissue underneath, not just a colour. A purpose-built chip captures all of it in a single frame, and on-device AI gives the surgeon both a normal-looking video feed to operate from and a live map of tissue oxygenation across the field of view. The result is real-time, objective measurement layered on top of the surgeon’s existing view, complementing tools like ICG rather than replacing them, and giving surgeons an additional source of confidence in moments where they currently have to rely on instinct.
The reason this matters beyond colorectal surgery is that the same questions surgeons currently answer largely by eye are everywhere in the operating room. Where does a tumor end and healthy tissue begin? Is this section of bowel, oesophagus, or stomach wall viable enough to heal? Surgeons can and do answer these questions today, but a meaningful share of the answer is informed guesswork, and the cost of guessing wrong is high. Each of these is a tissue physiology question hiding inside a visual one, and each becomes more answerable when the camera can measure physiology directly instead of leaving it to inference. The hardware does not change for each new question. The software does. And the chip is designed to fit into the vision systems already used in laparoscopic, robotic, and endoscopic surgery.
The bigger picture
The deeper reason we are excited is what this enables over the long term. Surgical robotics today is, for the most part, an extension of the surgeon’s hands, with the surgeon still making essentially every decision in the case. Over time, that will change. The next generation of surgical systems will assist more, standardize more, and eventually share more of the cognitive load. That future will run less on guesswork and more on objective, reproducible measurement of what the tissue is actually doing, the kind of ground truth that lets a system, and the surgeon working with it, agree on what they are looking at. Better robots, better algorithms, better autonomous capabilities over time, all of them eventually depend on something that can see the patient as more than a coloured surface. That is the sensing layer Hypervision is building.
Hypervision came out of King’s College London, where the CEO Michael Ebner and his co-founders built the core technology. The chairman, Martin Frost, founded CMR Surgical, one of the most successful surgical robotics companies in the world. The newest member of the board is Rick Mangat, who founded NOVADAQ and pioneered intraoperative ICG fluorescence imaging.
The hardest companies to build (and invest in) are the ones that have to be right about the science, the engineering, and the clinical reality all at once. Michael and the team are. We have huge respect for them and are thrilled to be on this journey with them.
P.S.: Reach out to Felix or to Michael if you want to learn more, partner with HVS or join the team! This is a scientific rocketship.
Speak soon,
Lucas
 | A guest post by
|