June 20, 2013
Diagnosing Heart Disease With Video Of Head Movements
Lee Rannals for redOrbit.com — Your Universe Online
A new way to diagnose cardiac disease by analyzing miniature head movements with digital video is being developed.
MIT researchers have created an algorithm that can accurately measure the heart rates of people through digital video by analyzing imperceptibly small head movements from the rush of blood caused by the heart's contractions. This algorithm was able to give pulse measurements that were consistently within a few beats per minute of those produced by electrocardiograms (EKGs).
“From a medical perspective, I think that the long-term utility is going to be in applications beyond just pulse measurement," said John Guttag, the Dugald C. Jackson Professor of Electrical Engineering and Computer Science and director of MIT's Data-Driven Medicine Group.
A video-based pulse-measurement system would be useful for monitoring newborns or the elderly because this group has sensitive skin that could be damaged by EKG leads. Guttag added that the team's technique could measure cardiac output, or the volume of blood pumped by the heart.
“I think this should be viewed as proof of concept," Guttag says. “It opens up a lot of potential flexibility."
The team's algorithm first uses standard face recognition to distinguish the patient's head from the rest of the image. After this, it randomly selects 500 to 1,000 distinct points, clustered around the patient's mouth and nose, and tracks the movement from frame to frame. The system then filters out any frame-to-frame movements caused by regular breathing and gradual changes in posture. Finally, the algorithm decomposes the resulting signals into several constituent signals, one of which is selected to be the most regular and falls within the typical frequency band of the human pulse.
Guttag and colleagues would be able to improve the accuracy of the algorithm by combining it with other video-analysis techniques.
Quanta Research Cambridge (QRCLab) helped fund the research due to its interest in developing a commercial system for monitoring newborns being transferred by ambulance between hospitals.
“We had these videos of these patients and we were looking for ways of learning more about the infants based on the video signal," said Stephen Lewin-Berlin, managing director at Quanta Research Cambridge. “Their skin is very sensitive, so probes that attach to the skin that are not problematic for adult patients can actually cause damage to the patient."
He said it is still an open question in understanding how some of the underlying biological functions are translated into measurable signals.
“So I think there are still interesting opportunities in the medical domain to continue to explore what the different changes in the video signal mean and how they relate back to the underlying biological function," Lewin-Berlin said.
Stanford University researchers have also developed a monitor with newborns in mind. They wrote in the journal Nature Communications in May that they created a wearable heart monitor that is thinner than a dollar bill. The team is now working on making the flexible monitor system wireless so doctors would be able to receive a patient's minute-by-minute heart status through a smartphone.