Past Projects

Wearable Health Monitoring

 


Recent Past Members

    Philip A. Shaltis, Ph.D. in Mechanical Engineering, MIT

    heart bullet Eric R. Wade, Ph.D. in Mechanical Engineering, MIT

    Levi B. Wood, Ph.D. Candidate in Mechanical Engineering, MIT

 


Recent Projects

    Adaptive Motion Artifact Reduction for Wearable Health Monitoring Sensors Using MEMS AccelerometersMotion Corrupted Wearable Signals


Wearable medical sensors are expected to revolutionize home health monitoring and sports medicine. The photoplethysmogram (PPG) sensor, in particular, contains rich information about heart pulsation and blood oxygen saturation as well as breath rate. More recently, the PPG has even been used for estimating arterial blood pressure. 


However, the physiologic information found in a PPG signal may be so significantly corrupted by non-physiologic artifacts that the measured signal can no longer be used for any medical assessment, as shown in the top figure to the right.


It is extremely difficult, or even impossible, to eliminate motion artifact induction into the PPG sensor signal. It is very difficult to eliminate contact disturbances/movements between the sensor and the skin. Even if it were possible to eliminate sensor contact artifacts, there is additional motion artifact due to the acceleration induced blood flow in the peripheral arteries.


Since it is not possible to eliminate motion artifact induction, we have instead explored a framework for removing motion artifact from a corrupted sensor signal.


As shown in the middle figure on the right, we equip a PPG sensor, in particular a PPG finger-ring senor, with a MEMS accelerometer, to measure sensor and arm motions. We then use the acceleration information with the corrupted PPG signal measurement in an adaptive noise cancellation framework to remove the portion of the PPG signal that is correlated with the motion. This yields a recovered signal that should be free of motion artifact.


We have found that, by using Widrow’s Active Noise Cancellation framework and a Laguerre series expansion to model the motion-to-artifact dynamics, we are able to consistently obtain results that are able to capture wave form timing and amplitude, as shown in the bottom figure on the right.




Related Publications

 L.B. Wood,  H.H Asada, “Low Variance Adaptive Filter for Cancelling Motion Artifact in Wearable Photoplethysmogram Sensor Signals,” Proceedings of IEEE Engineering in Medicine and Biology Conference, Sept. 2007. (Poster Presented at Conference)

 L.B. Wood,  H.H Asada, “Active Motion Artifact Reduction for Wearable Sensors Using Laguerre Expansion and Signal Separation,” Proceedings of IEEE Engineering in Medicine and Biology Conference, pp 3571-3574, Sept. 2005.

P.T. Gibbs, L.B. Wood,  H.H Asada, ““Active Motion Artifact Cancellation for Wearable Health Monitoring Sensors using MEMS Accelerometers,” SPIE Smart Structures and Materials 2005, Vol. 5765 pp. 811-819, May 2005. 

 


Revised: September 23, 2007.