THIS CONTENT IS BROUGHT TO YOU BY NTNU Norwegian University of Science and Technology - read more

Researcher Andreas Østvik demonstrates the equipment that uses artificial intelligence to harvest experiences from previous patients. This way, doctors can make decisions based on potentially thousands of similar examinations.

Artificial intelligence can be used to save even more cardiovascular patients

If anything goes wrong with your heart, it is critical to get the right treatment quickly. 

Cardiovascular diseases claim the most lives worldwide, accounting for 17.9 million deaths every single year, according to the World Health Organization (WHO). To put it in perspective, this is almost double the population of London.

Therefore, it is crucial that the examinations conducted are as precise as possible, in case you suddenly need to check your heart.

“One of the most important methods is the echocardiograph,” Bjørnar Grenne says. He is a senior consultant in the cardiac department at St. Olav’s Hospital, and an associate professor at NTNU.

Every year, a large number of patients are admitted to St. Olav’s Hospital to have their hearts checked. This includes people experiencing chest pains, individuals collapsing on the street, and patients requiring regular heart checks. 

Many of them lie down on a bench and are examined by a doctor who uses ultrasound as a tool to look inside the body.

“The heart is extremely complex and is very well hidden within the body. We don’t think about it being there, but it’s there all the same, beating up to 100,000 times a day, every day – in each and every one of us,” Grenne says. "There is a good reason why the heart is so well hidden, but because we cannot see it, this also makes it harder to examine. That’s why we need good ways of studying it.” 

Demands experience and time

WHO has determined that cardiovascular diseases accounted for 32 per cent of all global deaths in 2019

“It’s important to establish what is wrong with the heart at an early stage, so that people can quickly get the right treatment,” Grenne says. 

At the examination table, senior consultant Grenne presents the probe, which looks like a joystick, and guides it to display the heart of a volunteer from the research team on the ultrasound screen. 

“This gives us real-time images of the heart, which are essential if we are to make the correct diagnosis,” he says. 

The challenge is that you need a lot of experience in order to guide this probe correctly and get the best possible images of the heart. Analysing the images afterwards is also very time-consuming.

“We can take as many as 70–100 different images and videos of the heart during an examination. These must also be studied carefully afterwards by people with a great deal of experience in this field, which can easily take half an hour when done correctly,” Grenne says.  

AI is the doctor’s super assistant

This is where AI, or artificial intelligence, comes in as an excellent assistant. 

“Artificial intelligence can help Bjørnar and his colleagues guide the probe in the right direction and obtain the perfect image every time. AI can also analyse the images as soon as they pop up on the screen and help us to see what is wrong with the heart,” says Andreas Østvik, a researcher at SINTEF Digital and NTNU.

Through machine learning, researchers have fed information into a machine, where Grenne and his colleagues have defined the criteria that must be met to obtain the right cardiac images, and how these images should be interpreted. 

In this way, artificial intelligence is used to draw insights from previous patients' experiences, allowing doctors to make decisions based on potentially thousands of similar examinations.

During the process, the AI assistant shows a green or red light, so that the doctor knows whether the probe is angled correctly. When the images are correct, AI interprets them and automatically takes measurements of the heart. This typically includes measurements of the heart's size and its ability to contract.

Testing on patients

The development is well underway, and some patients have already received this as part of their treatment through the research project. However, strict rules apply in the field of medicine, so it may take some time before this treatment option is available in all hospitals in Norway.  

“We’re expecting it to be available in a few years. We have to test it first – patient safety is paramount, and we have to know that it works before it is made available everywhere,” the researcher says.

Grenne adds that the current way of examining the heart is very effective, but that it will obviously be tremendously helpful if AI could contribute as an assistant. 

“It saves us a lot of time and resources, which means that we can help even more people – which could then save more lives,” he says.

Powered by Labrador CMS