BUILT FOR INDUSTRY, ADOPTED BY THE HEART. AN ACCIDENTAL MEDICAL AI STORY

Before hospitals, before cardiology, before medical regulations and clinical trials, there was a team working with artificial intelligence in an entirely different world. Their daily reality was manufacturing: machines, processes, and industrial efficiency. What if this technology were used where it could matter most?

That was the starting point of the AI4CMR project (Artificial Intelligence for Cardiovascular Magnetic Resonance).

The project was spun out of Neadvance, a company with strong experience in industrial machine vision. For years, the team applied artificial intelligence to manufacturing problems, improving efficiency and quality control. Move from industrial AI to hospital wards, from algorithms to patients, and from an idea to a project with the ambition to make advanced heart imaging more accessible, safer, and faster for everyone.

“When you look at medical images, the value is different; you’re not just improving performance. You’re helping someone take care of another human being.”

Says António Murta, co-founder and CEO of AI4CMR.

That realisation became the seed of AI4CMR. António was one of the founders, together with Teresa Martins from Neadvance, and early on, by Prof. Nuno Sousa, whose role proved decisive. It was Prof. Sousa who helped select the two initial medical domains to explore: one in gastroenterology and the other in cardiac magnetic resonance (CMR).

For the first 18 months, the team explored both paths in parallel. They listened to clinicians, studied workflows, and tried to understand where AI could genuinely help. Finally, the decision to focus solely on cardiology was driven by realism rather than lack of ambition. “We simply didn’t have the means to support both projects,” António says. CMR became the clear choice.

The reasons were compelling. CMR is the gold standard for assessing heart function and structure, yet it is complex and time-consuming. Highly trained specialists are needed, and demand for the exam has been rising faster than the number of experts available to read and interpret it. This mismatch between supply and demand worsened during the pandemic and remains evident today. AI4CMR was created to help close that gap, not by replacing clinicians, but by supporting them in their work.

Building the right team was a gradual process, shaped by different needs at different moments. The technical foundations were already there, but medical and product vision were essential. Very early on, Vítor Hugo joined the company and played a critical role. He led product visioning, translated clinical needs into technical requirements, and helped guide development for several years.

Other contributions followed as the company evolved. Daniel Leite, now Commercial and Marketing Officer, joined much later, although, as António notes, “we talked about Daniel joining many years before he actually did.” For Daniel, the timing mattered. “After many decades in the industry, this project brought me back to something more meaningful,” he says. “It’s not just about improving margins, but about improving patient care.”

More recently, Manuel Monteiro joined the team, bringing deep experience in hospital IT management. His understanding of hospital systems and customer environments added another crucial layer as AI4CMR moved closer to large-scale adoption.

Like any entrepreneurial journey, this one was marked by obstacles. Regulatory barriers were among the most demanding. As a medical device, AI4CMR had to comply with strict rules in Europe and beyond. António admits:

“We climbed a big mountain, not just in terms of money, but in terms of pressure on the team.”

Data availability was another constant challenge. Medical AI depends on large, high-quality datasets, yet access is often slow and complex, particularly in Europe. “It’s always an issue in medical AI,” António says plainly.

Market adoption brought its own surprises. The original plan was to focus on algorithms and rely on existing PACS systems for visualisation. That proved unrealistic. Leading PACS tools were not designed for advanced CMR workflows, especially when contour editing and detailed analysis were required. The team faced a difficult decision: either accept limitations or pivot. They chose the latter, developing their own visualisation approach through a close partnership with Meddream. “It was a challenging moment,” António recalls, “but it made the solution much stronger.”

Funding was another strategic hurdle. AI4CMR operates in a highly specialised field, cardiology software, and “smart investors” with the right expertise are scarce in Europe. In this context, the European Innovation Council (EIC) grant played a vital role. “It was a bridge,” António says, “that helped us cross a very difficult chasm.”

So how did the team stay motivated through slow progress and setbacks? “Resilience, first of all,” António answers. “It’s a must in a start-up.” Motivation also came from early signals of product–market fit and support from key opinion leaders. Experts such as Tim Leiner from the Mayo Clinic and Nuno Bettencourt from Unilabs offered guidance and encouragement, reinforcing the sense that the work mattered.

Support from partners and stakeholders also made a difference, from the EIC and the EIC Medical AI Community, including Federica Zanca, the EIC Programme Manager for AI in healthcare, to trusted industry partners such as Vytautas from Meddream and Carlos Cardoso from Sectra. These relationships helped the team feel less alone in a demanding journey.

Looking ahead, the mission is clear: to help democratise access to CMR, a critical exam in cardiology guidelines. By simplifying and speeding up analysis, AI4CMR aims not only to reduce the gap between supply and demand but also to increase safety, for patients and for experts. Automating reports and eliminating risky copy-and-paste of quantitative data is a small change with potentially life-saving consequences.

For António, the impact is deeply personal. “Helping improve medical practice is not just work,” he says. “It’s a way of living and defending life.” That belief continues to drive AI4CMR forward and stands as a reminder that behind every line of code, there is a human reason to keep going.

Photo by benjamin lehman on Unsplash