Normal webcams will diagnose cardiac problems

A web camera can analyze a person’s face and diagnose whether that person is experiencing atrial fibrillation, a treatable but potentially dangerous heart condition.

January 2015

A perfectly normal webcam connected with sophisticated software is capable of detecting atrial fibrillation (AF) with roughly the same accurately achieved by a standard electrocardiogram test (ECG). ECG tests are carried out on people with certain precise symptoms suggesting cardiac problems, and they involve the use of special instruments attached to the person’s body and the presence and skills of a doctor.  This new facial technology, on the contrary, can detect atrial fibrillation in people with no special symptoms, without needing physical contact, simply by analyzing video footage of a person’s face and detecting subtle shifts in skin color which indicate changes in blood flow.

There are approximately 3.2 million people with AF in the U.S., and an estimated 30 million people in the world. But the real problem is that 30 percent of people with AF are not aware that they have it. This new technology is potentially extremely important, because currently there’s no way to diagnose the condition without consulting a doctor and carrying out tests. AF is a progressive condition, so early diagnosis is vital, especially since the long-term implications are serious and it can end up causing strokes or heart failure.

The technology in question has been developed in a partnership between the University of Rochester School of Medicine and Dentistry, on the one hand, and Xerox, on the other. It employs a software algorithm which scans a digital video of a face that allows it to register changes in skin color which are imperceptible to the naked eye.  All that is required of the subject is that they remain still for 15 seconds. 

Sensors in digital cameras are designed to record three colors: red, green, and blue. Hemoglobin – a component of blood – “absorbs” more of the green spectrum of light and this subtle change can be detected by the camera’s sensor. Fortunately, the face is the ideal place to detect this phenomenon, because facial skin is thinner than that of other parts of the body and blood vessels are closer to the surface.

Experiments have clearly demonstrated that the observed color changes do effectively correspond with the subject’s heart rate.

The scan correctly identified the presence or absence of atrial fibrillation 80 percent of the time, with an error rate of 20 percent. This average score was not dissimilar to that of automated ECG measurements, which had an error rate between 17 percent and 29 percent of the time.
The research team is convinced that as the research develops – along with the application of image stabilizing technology and the ongoing improvement in the resolution of cameras – the error rate will get progressively lower. 

Like many other personal health technologies that have emerged in recent years concerning numerous potentially hazardous medical conditions, the authors see this method as an important way to diagnose or monitor people who risk atrial fibrillation and to alert them and/or their physicians when the condition is detected. The contactless nature of the technology and the proliferation of web cameras could even eventually allow the screening to occur without interrupting the user. For example, the program could simply run in the background while someone is reading emails on their tablet, computer, or smart phone.

Among the aspects which will need further research and fine tuning , it’s interesting to note that so far the experiments have been conducted only on people of Caucasian racial origin, so the technology will also have to be tested on population groups with other kinds of skin coloring.