Think of video analytics and you may picture rows of crime analysts monitoring multiple screens on a giant wall. But algorithms designed to spot unusual activities and predict potential hot spots are now moving beyond surveillance. The medical field, for example, is at the forefront of embracing similar video and image analytics technologies.
The Industrial Internet Meets Healthcare
It’s easy to see why image analytics algorithms make sense to healthcare providers. The medical field has been a pioneer in capturing rich imaging and video information and building it into databases. And as healthcare meets the Industrial Internet, doctors are using more and more sensor-fitted medical devices that can seamlessly communicate with each other. These create a huge stream of video and image data in every step of diagnosis, treatment, and recovery. Thanks to big data technologies, varied information contained in this pixel data is being crunched and connected to genetic data, patient history, labs, and social history.
Research activities are underway internationally to come up with different kinds of algorithms to crunch this imaging data. The early results are promising to change the health landscape dramatically.
Predicting Genetic Disorders
Image analytics can play a significant role in predicting certain disorders. Researchers from Oxford University have developed a computer algorithm that can analyze photographs and diagnose which children have certain genetic disorders. The researchers say their computer program recognizes facial features in photos and compares them with facial structures resembling various conditions — including Down syndrome, Angelman syndrome, or progeria — and returns potential matches ranked by likelihood. Computational analysis of the shape of faces using 3D imaging has also been carried out to analyze conditions such as fetal alcohol syndrome, schizophrenia, and Alzheimer’s.
Monitor Vitals with Webcams
Hospitals have used virtual patient observation systems to monitor patients at risk for falls, in other personal harm situations, and when confused or agitated. Video analytics now make it possible to monitor vitals remotely, too. Engineers at Xerox Corporation's research centers in Bangalore, India, and Webster, New York, are testing the feasibility of video sensing combined with data analytics to track the status of patients with chronic conditions. Light emitted from a camera penetrates into the patient's skin, allowing information, like redness of skin, to be captured. Using patented software algorithms, researchers have been able to convert data collected by these cameras into vital signs. Researchers at the University of Rochester are also exploring the use of cameras to detect atrial fibrillation, a rapid quickening of the heart that boosts a patient's risk of stroke.
Scan and 3D Print Organs
Algorithms not only use images in patient records to diagnose and predict but also combine three dimensional imaging capabilities of computer tomography (CT) with 3D printing to produce exact scale models of various organs. CT scans take a rotating image of successive thin slices of the body; these are then reconstructed with software and displayed on a computer monitor. Researchers at Notre Dame University in Indiana used an Albira CT scanner to return images and associated data with sufficient resolution for the 3-D printers to produce at least some models. Similarly, surgeons in the U.K. performed facial reconstruction on a motorcycle crash victim and hip implant for a patient suffering from a rare degenerative bone disorder called Recklinghausen’s disease.
Many more video and image analytics projects, currently in research labs, are promising to open new opportunities, as healthcare taps into data science and logs onto the Industrial Internet.