How a machine learning driven solution is making cardiac MRI more efficient, more consistent, and less costly


“My mother used to be a secretary in the dialysis ward of a hospital,” says Itamar Kandel, CEO of Californian MedTech startup HeartVista. “Every day after school, I’d make my way to her workplace. I used to run between machines and speak with the medical staff. So the smell of hospitals’ antiseptics and seeing linoleum on the floor formed a large part of my childhood memories. The unusual exposure made me fall in love with medicine and I always thought I would one day become a doctor.

“I eventually ended up studying computer science instead,” Kandel continues, “because I developed a passion for the field as I grew up, but I have never forgotten the love of my childhood. I wanted to be able to combine both technology and biology in a way where they will have a dramatic impact on people’s lives. That explains why I am in biotech.”

However, heading a MedTech company was a somewhat unexpected opportunity. Kandel recalls receiving a call from a Silicon Valley recruiter while in the hills of Kandy in Sri Lanka, asking for his interest to be the CEO of HeartVista. “I was on my sabbatical year, backpacking through Southeast Asia. I remembered telling the recruiter I wouldn’t be back for another nine months and if he can hold the position for that long, that would be great. Otherwise, let’s keep in touch and see what happens.” Kandel thought he’d missed the opportunity. But the same recruiter called him again on the day he returned back to the US. “We started the conversation again as he told me how the role went away and came back. At that moment, I thought the whole thing was meant to be.”

Headquartered in Los Altos, California, HeartVista leverages machine learning to facilitate the process of cardiac MRI; making it faster, easier, more consistent while minimizing the need for breath-holding. Specifically, a comprehensive ischemia exam can be performed in less than 15 minutes.

The HeartVista software also automatically prescribes the standard cardiac views in as little as 10 seconds while patients are breathing freely. The detection algorithm is incorporated into the solution’s autonomous protocol to signal when the image quality is below the acceptable range, prompting the MRI technologists to reacquire the questioned images if needed.

“A study published in the journal Circulation: Cardiovascular in September 2019 revealed that cardiac MRI has become the new gold standard to diagnose many related diseases,” Kandel says. “Past studies found machine learning automated cardiac MRI is superior in terms of speed and its ability to evaluate cardiovascular structures and functions and to quantitatively assess blood flow and perfusion without the use of radiation or the need for invasive catheterization.” Despite that, cardiac MRI still only accounts for 8% of cardiac scans.

Creating diagnostic-quality CMR images requires up to 300 different runtime decisions to perfect the signal output. MR technicians executing the exams must be well-versed in the anatomy of the heart and the physics of the machines to make the decisions. Often, hiring, training, and retraining these highly paid CMR technicians is daunting. Even with a trained technician, it would need up to 90 minutes to perform a single cardiac MRI examination, making the process costly and the idea of having a CMR might be put off in busier medical centers where there are many more patients needing the scanner for a shorter time.

As such, the HeartVista solution aims to take control by integrating with existing MRI machines and automate cardiac MRI from a complex multiple clicks session to one which involves only one click. “That means a patient comes into the room, the clinician situates them in the machine and clicks on the machine so that it takes over and runs the exam from start to the end,” Kandel explains.

The technology originated from Stanford University’s Magnetic Resonance Systems Research Laboratory (MRSRL), where researchers spent decades developing advanced and real-time MRI applications and other tools that assist physicians to accurately diagnose cardiovascular diseases.

Unlike many startup owners, Kandel is not governing a solution which he could call his own, and his family expressed doubts that he would be ideally suited for the role. “I admit that being an entrepreneur is not the easiest path one can take,” Kandel says. “My father was a union leader for 40-odd years. When he looked at what happens in Silicon Valley, he told me I’ve got to be comfortable with the cycle of success and failure and moving fast with uncertainties.

“But I believe vulnerability is not a weakness, it’s a sort of strength,” Kandel adds. “And when you present yourself fully without all the pretences, without the image that you’re trying to portray, I think people react well to that. I think empathy and creating real connections is the way to go.”

It’s a philosophy that’s starting to pay dividends. Under Kendal’s leadership, HeartVista has raised $8.65M in Series A funding led by Khosla Ventures, Jeff Rothschild, Leslie Ventures, Open Field Capital and other investors. It’s planning to use the money to fuel new musculoskeletal and neural products, international expansion, and deepen strategic alliances with some of the world’s leading cardiology centers.