BME Graduate Student Exchange Seminar

Event Date: April 26, 2013
Time: 9:30 a.m.
Location: MJIS 1001, WL campus
Contact Name: Steven Lee
Contact Email:
***SPECIAL SEMINAR: BME Graduate Student Exchange with Northwestern University *** "Intent recognition for powered lower limb prostheses using neural interfaces" presented by Aaron Young, Doctoral Candidate Biomedical Engineering, Northwestern University, Rehabilitation Institute of Chicago, Northwestern University (Advisers: Todd Kuiken, M.D., Ph.D and Levi Hargrove, Ph.D.)

Abstract: New powered lower limb prostheses are being developed that allow amputees to better traverse multiple types of terrain. Current devices are not capable of transitioning automatically and seamlessly between locomotion modes such as sitting, standing, level walking, stairs and slopes. The focus of my study was to enhance intent recognition interfaces with an electromyographic (EMG) interfaces. Additionally, novel bayesian algorithms were applied to incorporate the time history information of both mechanosensory and electromyographic signals to further improve intent recognition accuracy. Five transfemoral amputees walked on a powered knee and ankle prosthesis using the intent recognition interface, and it was shown that incorporating EMG and time history information reduced the misclassification rate from 6% to 2% (or a two-thirds reduction). These results suggest that the application of myoelectric and pattern recognition technology can consistently and accurately predict amputees' intent for everyday activities.

Biography: Aaron Young received his bachelor’s degree in biomedical engineering at Purdue University in 2009 and completed his master's in biomedical engineering in 2011 at Northwestern University where he is now a doctoral candidate in the Center for Bionic Medicine under the guidance of Dr. Todd Kuiken and Dr. Levi Hargrove. His primary research interests are biological signal processing, machine learning, upper and lower limb prostheses, myoeletric control and amputee intent recognition.