Facial thermal imaging used with AI can detect coronary artery disease, study says

Facial thermal imaging "can identify areas of abnormal blood circulation and inflammation activity through the measurement of skin temperature patterns," researchers wrote in their findings. Photo by Mart Production/Pexels
1 of 3 | Facial thermal imaging "can identify areas of abnormal blood circulation and inflammation activity through the measurement of skin temperature patterns," researchers wrote in their findings. Photo by Mart Production/Pexels

NEW YORK, June 4 (UPI) -- Facial thermal imaging, combined with artificial intelligence, can correctly detect coronary artery disease better than traditional methods, new research from China suggests.

The findings were published Tuesday in BMJ Health & Care Informatics.


"This non-invasive, non-contact technique may hold promise for coronary artery disease screening in the community," the study's corresponding author, Dr. Zhe Zheng, chief doctor in the National Center for Cardiovascular Diseases at Fuwai Hospital in Beijing, told UPI via email.

It's the world's largest cardiovascular center, according to the World Health Organization.

Facial thermal imaging "can identify areas of abnormal blood circulation and inflammation activity through the measurement of skin temperature patterns," the researchers wrote in their findings.

However, they cautioned that validating this technique's effectiveness on more substantial and ethnically diverse groups of patients would be necessary before its adoption in clinical practice.


Present guidelines for diagnosing coronary heart disease depend on the probability assessment of risk factors that aren't entirely accurate or widely applicable, so the researchers called for investigating more precise detection tools.

The results, though, were most encouraging.

During testing, "our algorithm had superior performance compared to the current guideline-recommended clinical tool for coronary artery disease screening," Zheng said.

Supplementing information about risk factors with other current methods is often time-consuming and invasive. In addition to blood work, clinicians frequently order an electrocardiogram to record the heart's electrical signals or an angiogram, which uses imaging to visualize blood flow through the vessels or heart, the researchers said.

By contrast, thermal imaging is noninvasive. It captures temperature distribution and variations on the object's surface by detecting infrared radiation.

"The advent of machine learning technology to extract, process and integrate complex information has shown impressive capability in harnessing myriad imaging information for various disease predictions," the researchers noted.

They explored the feasibility of using thermal imaging plus AI to accurately detect the presence of coronary artery disease without resorting to invasive, time-consuming techniques.

In the 460 people with suspected heart disease in the study, the average age was 58, and 126 (27.4%) of them were women.


The study captured thermal images of participants' faces before confirmatory examinations to develop and validate an AI-assisted imaging model for detecting coronary artery disease.

Researchers confirmed that 322 participants (70%) had coronary artery disease. They tended to be older and more likely to be men. They also tended to have lifestyle, clinical and biochemical risk factors, as well as higher use of preventive medications.

The combined approach of thermal imaging and artificial intelligence was about 13% more effective in detecting coronary artery disease than the pre-test risk assessment involving conventional risk factors and clinical signs and symptoms, researchers said.

Among the three most significant thermal indicators, the most influential was the overall left-right temperature difference of the face, followed by the maximal facial temperature and average facial temperature.

In particular, the average temperature of the left jaw region was the strongest detection feature, followed by the temperature range of the right eye region and the left-right temperature difference of the left temple regions.

This approach also effectively identified traditional risk factors for coronary artery disease: high cholesterol, male sex, smoking, excess weight, fasting blood glucose and inflammation indicators.

As for limitations, the researchers acknowledged the relatively small sample size in their investigation and the fact that it was conducted at only one hospital. The participants had all received referrals for tests to confirm suspected heart disease.


The study only enrolled those of Chinese origin, so it's unknown how well this screening method would work on people with different skin pigment, said Dr. David Maron, director of preventive cardiology at Stanford University School of Medicine in Palo Alto, Calif. He was not involved in the study.

However, "to my knowledge, this is the first effort combining facial skin temperature patterns and AI to predict the presence of coronary artery narrowing," Maron said.

"This test performed better than conventional risk factors, such as high cholesterol and high blood pressure, to assess a person's risk for heart disease," he added, noting that "temperature may reflect an individual's state of inflammation" -- a cause of coronary artery disease.

Significant research into facial temperature needs to occur before this tool can be integrated into patient care, but if validated, it may become an efficient, cost-effective way to screen many patients for cardiac risk, said Dr. Mitchell Weinberg, chair of the department of cardiology at Northwell Health's Staten Island University Hospital in Staten Island, N.Y.

"Growing patient volume, cardiac testing and the increasing complexity of patient disease place even greater demands on today's cardiologist," Weinberg said. "Artificial intelligence might emerge as the future cardiologist's most valuable clinical partner."


While this technology will probably represent the next chapter in clinical medicine, "we're many years away from machines taking away decision-making from clinicians, if ever," said Dr. Shahbaz Malik, director of the cardiac catheterization laboratory at Nebraska Medical Center in Omaha.

Even so, in the near future, Malik said he expects "machine-based learning will serve to be an important tool at the disposal of clinicians diagnosing and treating patients."

"This kind of study is a wonderful demonstration of using AI for patient-centric, noninvasive diagnostic purposes -- very innovative," said Dr. Jennifer Avari Silva, chair of the Health Care Innovation Council at the American College of Cardiology, an organization of cardiovascular specialists.

It's "a solid first step toward a clinically and commercially viable tool" for diagnosing coronary artery disease, said Silva, a professor of pediatrics and biomedical engineering at Washington University in St. Louis.

"Cardiology is a minimally invasive field with technology quite suited to AI analysis."

Latest Headlines