There are plenty of drugs that lower cholesterol, yet heart disease is still the leading cause of death in the US. Elucid, a Boston medical software startup, hopes that artificial intelligence can help cardiologists spot early signs of heart disease and make recommendations about how best to prevent heart attacks and strokes.
The company has developed software that uses machine learning to analyze heart scans to identify and characterize plaques in the coronary arteries, which can cause deadly clots if they break free.
The software is cleared for use by the US Food and Drug Administration, and Elucid said Wednesday that it has raised $27 million in series B funding to begin commercializing it. Chief executive Blake Richards said he believes the company’s approach “has the potential to change the way we treat cardiovascular disease and save millions of lives a year.”
Elucid’s origins go back to 2008, when the company was founded under a different name by Andrew Buckler, an engineer and computer scientist with imaging technology experience at Philips, Hewlett-Packard, and Eastman Kodak.
During his years at those companies, Buckler noticed that although medical imaging was improving, heart scans were still a subjective enterprise left to an expert’s interpretation. He thought there was room to improve, so he began his own company to develop software that could help cardiologists better assess the risks of heart attacks and strokes.
Work was slow going at first, and much of the company’s first decade was supported by government grants and a consulting business. Elucid finally raised an $8 million series A fund last year to finish developing the software. It’s designed to help cardiologists glean more information from a CT angiography, a heart scan that doctors sometimes recommend when a patient complains of chest pain.
“We go beyond what is visually conspicuous, truly complementing the visual intelligence of a physician,” Buckler said. The goal isn’t to replace doctors, but rather to arm them with a tool that highlights things they cannot see with the naked eye.
Richards said that the company’s software can determine the blood flow through coronary arteries and predict which patients will be good candidates for a stent. Another company, HeartFlow in Mountain View, California, sells similar software. But Elucid’s product can also assess the stability of a plaque and predict how likely it is to break away from the artery wall, Richard said.
“That can’t be done by any other company in any way right now,” said Dr. Robert Pelberg, a cardiac imaging specialist and clinical cardiologist at the Christ Hospital in Cincinnati, Ohio. He is also one of a dozen doctors and researchers on Elucid’s large scientific advisory board.
“Plaque characterization is the holy grail of imaging, and Elucid has solved that problem,” Pelberg said.
The software was based on dissected coronary arteries from patients that were analyzed by pathologists. Those laboratory studies were used to train the software to assess the arteries and plaques from non-invasive heart scans, “as though a pathologist had it under a microscope,” Richards said.
Pelberg believes that Elucid’s ability to characterize plaques in the vessels of a beating heart could provide a new way for drug developers to test plaque-busting and cholesterol-lowering drugs. In fact, Richards said that his company already has contracts with some pharmaceutical firms to use the software.
Although statins are widely prescribed, they are not necessarily the best drug for everyone. Alternative drugs are not used as often because they are more expensive, have more side effects, and there isn’t a good test to predict who will most benefit from them. Elucid is working on improving its software to predict which class of cholesterol-lowering drug would best help a patient.
“It is not pie in the sky,” Pelberg said. “It has very real potential, but it is a very long way to reality.”
Although some academic medical centers are already using Elucid’s software, the startup’s new funding will help it expand its commercial team to get the software in more hospitals, which would likely pay for it on a per-use basis.
Crucially, the company said its software explains its diagnoses and recommendations. “Everything we do is interpretable by the physician,” Richards said. “We are not some black box where the information that comes out is not explainable.”
Based on the firm’s competition, Buckler anticipates that Elucid could be reimbursed more than $1,000 per patient.
“We think that in the future every CT angiography should run through an analysis such as ours,” Richards said.
Elucid has about 40 employees, and Richard expects the headcount to increase, but he wouldn’t make specific projections about the company’s growth or potential sales.