Computer Vision Clinical Monitoring
NREC and Columbia University developed a computer vision algorithm to monitor patients in clinical trials for spinal muscular atrophy (SMA) therapies. The vision system can reduce the cost of clinical trials for drug therapies and speed up their approval by the FDA.
Application
In Spinal Muscular Atrophy (SMA), a child’s motor neurons gradually lose their ability to carry messages from the spinal cord to the muscles. This takes away the patient’s muscle strength and breathing ability; in severe cases, it can cause death.
Patients in clinical trials for SMA therapies must be admitted to a hospital or monitored at home by a clinician, both of which are expensive. Because it affects movement, SMA is a good candidate for computer vision. Patients can be observed in their homes during clinical trials by a vision system that monitors their breathing and other muscular movements. The vision data would show how a therapy affects the patient’s muscle tone, respiration, and other markers for SMA.
Automated home-based monitoring is much less costly than hospital admissions or clinician visits, lowering the financial barrier for approving new therapies. It is also more comfortable and less disruptive for patients and their families.
Capabilities
The pilot study examined whether respiration data gathered by a computer vision system correlated with respiration data recorded by body sensors and clinical observation.
Patients in the study wore a sensor vest (or life shirt) that recorded their respiration. The vision system tracked the three-dimensional movement of markers (or fiducials) on the life shirt. Additionally, it tracked the movement of the patient’s wrists and ankles, which also can indicate the progress of the disease. The cameras used infrared light, which is invisible to humans. This allowed patients to be monitored both day and night without disturbing their sleep. The system compared the computer vision estimates of the patient’s respiration with respiration data from the life shirt. It also compared the computer vision system’s observation of limb movements with those gathered by human clinicians. Movement and respiration data from the vision system correlated well with the data from the life shirts and clinical observations. This shows that computer vision monitoring can be used in clinical trials of SMA therapies.
