AngioWave Imaging - the four-person Boston outfit teaching old X-ray machines a new trick.
Company Profile
AngioWave
AI that finds the blood vessels your angiogram was hiding all along.
AI that finds the blood vessels your angiogram was hiding all along.
An interventional cardiologist stands over a patient, a coronary artery somewhere on the monitor knotted shut. The angiogram is there, looping in grainy black and white, and the path through the blockage is somewhere in that flicker - if only it would hold still long enough to be seen. This is where AngioWave Imaging works: not inventing a new scanner, but quietly improving the one already in the room.
AngioWave is a small Massachusetts company - roughly four people on payroll, an outsized board of cardiologists, and one stubborn idea. Its software, AngioWaveNet, takes the ordinary angiogram a hospital already produces and runs it through a deep neural network. What comes out the other side is the same heart, the same vessels, only legible. The company calls the technique STEP. Cardiologists, less poetically, call it "wait, I can see that now."
Coronary artery disease kills more people than anything else on earth. The standard way to look inside a beating heart is fluoroscopic angiography - X-ray dye, a moving image, a clinician's trained eye. It works. It has worked for decades. The trouble is that the rawest, most decisive details - a thread-thin collateral vessel, the true entry point of a chronic total occlusion - can hide in the noise, the motion, and the milliseconds.
The industry's usual answer is to buy a better machine. New hardware, new install, new capital budget, new line in a hospital's already strained finances. AngioWave noticed something almost inconvenient: most of the missing information was already captured in the existing image. It just hadn't been read properly.
The unlikely origin: the core mathematics came from pediatric neurosurgery. Co-founder William Butler, MD, a pediatric neurosurgeon at Massachusetts General Hospital, had developed wavelet and shearlet transform methods to study delicate neurovascular conditions in children. Vessels are vessels, it turns out, whether they sit in a skull or a chest. The same math that clarified a child's brain could clarify an adult's coronary arteries.
The other half of the bet is Aram T. Salzman, the CEO, a repeat founder who had previously started NovoBiotic Pharma and CRA Health. Butler brought the clinical insight and the patents; Salzman brought the unglamorous discipline of turning a clever algorithm into a cleared medical product. They wagered that a software layer - not a steel box - could become the thing every angiogram passes through.
Serial healthcare founder. Previously launched NovoBiotic Pharma and CRA Health before betting on imaging.
Pediatric neurosurgeon at Mass General and inventor of the wavelet-based imaging math behind AngioWaveNet.
Two founders, nine patents, and a whiteboard that started in a children's hospital. The heart was an afterthought - in the best way.
Most image filters sharpen a single frame. AngioWaveNet is built differently. It is a spatio-temporal, convolutional deep neural network - which is a mouthful that means it watches how vessels move across an entire cine, interpreting motion, structure and time together. A flickering collateral that vanishes for three frames and reappears is exactly the kind of thing a human eye loses and a temporal model keeps.
Practically, it asks nothing new of the hospital. The software ingests standard DICOM XA cines from any fluoroscopy system, does its post-processing, and hands back a clearer version. No new install in the cath lab, no rip-and-replace. Clinicians have reported better visualization of vessel structures, faster procedural decisions, and more confidence navigating complex anatomy. The same applicability reaches beyond cardiology - peripheral artery disease, trauma, and neurology and stroke imaging.
Six lines, one stubborn idea. Note how long the gap is between "good math" and "thing a hospital can buy."
Promising medical AI is common; cleared medical AI is rarer. AngioWave has FDA 510(k) clearance, nine issued US patents with more pending, and protection extending across the UK, Canada, the EU, Japan, China, Korea and Australia. The intellectual-property map is, frankly, larger than the company's headcount - which tells you where the value is concentrated.
The strongest evidence is a single case. In the first clinical use of AI-based STEP enhancement, the software surfaced a crossing pathway and a collateral vessel during a chronic total occlusion procedure - details that simply weren't visible on the unprocessed angiogram. One case is not a trial. But for a cardiologist standing over a blocked artery, it is the difference between guessing and seeing.
A lopsided chart on purpose: AngioWave is a heavyweight in IP riding on a featherweight payroll. That gap is the whole strategy.
AngioWave's ambition is deliberately narrow and quietly large. It does not want to sell hospitals a new imaging system. It wants to be the post-processing layer that every angiogram passes through, on whatever hardware already happens to be bolted to the ceiling. Software scales in a way steel never will, and a coronary blockage in Boston looks much like one in Bangalore.
The supporting cast underwrites the seriousness: a board and advisory roster that includes cardiology figures like Deepak Bhatt, MD and Paul Ridker, MD. For a company you could fit in a single conference room, the gravity of the names around the table is its own kind of signal.
Return to that cardiologist, that knotted artery, that flickering loop. The bet AngioWave is making is that the most valuable upgrade in the room is not a machine but a better way to read the one already there. If they are right, the next decade of angiography improves not by replacing equipment but by adding a thin, invisible layer of intelligence between the X-ray and the eye.
Now the angiogram loops again. Only this time the path through the blockage holds still long enough to be seen, the collateral that hid for three frames stays put, and the decision that took a guess takes a look instead. AngioWave didn't buy the cardiologist a new heart-imaging machine. It just made the old one tell the truth.