Even university researchers play Xbox from time to time. Not all mechanical engineers, however, have the idea to use the popular gaming console’s camera to assist doctors in clinical applications.
McGill University postdoctoral fellow Farnood Gholami employed the Microsoft Kinect, the Xbox’s depth-sensing camera, to assess stepping motion and identify gait characteristics indicative of multiple sclerosis (MS), a disease of the central nervous system. The imaging technology could improve physicians’ traditional ways of treating people who have MS.
MS Gait Patterns
Gholami and a team of researchers used the Kinect device, along with developed software, to detect the differences in the gait of MS patients compared to those without the disease.
By locating a subject’s joints, including the hip and the knee, the camera plotted the body parts’ position in space and detected the variations in step and pose.
To distinguish an abnormal gait from a normal one, and potentially quantify the degree of abnormality, the researchers’ software defined gait indices related to strides, joint angles, and walking patterns.
The team found that specific gait characteristics, measured with the Kinect camera and analyzed with Gholami’s algorithms, were common to people with MS.
A multiple sclerosis patient’s walk often varies in speed, step, or hip angle. Because individuals with MS often struggle with balance, for example, they may have a wider step width and shorter step length to maintain stability. Many with the disease also use double-support when walking: keeping both feet in contact with the ground as opposed to a single, leg-swinging stride.
Joint angles, another important indicator of the disease, determine a hip and knee’s range of motion. Using hip- and knee-joint angle information, a new index from the McGill team captured the general gait pattern of a subject and its deviation from a healthy one.
Settling the Score
Previous assessment of gait abnormality – and the severity of walking deviation – relied upon the pure observation of the physician. A clinician would watch a person’s walking pattern and assign clinical scores – an often subjective assessment, according to Gholami.
“Different physicians may come up with different clinical scores with the same subject,” said Gholami. “We had this idea to make it more standard and systematic, and use technology for gait assessment.”
Collaborating with the Montreal Neurological Institute and Hospital at McGill University, Gholami and his team recruited ten MS patients, and ten control subjects without multiple sclerosis. Using the Kinect device, the researchers stored gait data for each individual and quantified walking patterns.
Based on statistical analysis and mathematical models, Gholami created the computer algorithm to determine if a person walking in front of the camera had a gait irregularity or not. The software predicted a gait score, the level of step deviation.
The MS patients had previously been assessed for gait characteristics using traditional clinician observations. The clinical measurements matched favorably with the determinations made by the algorithm.
“We compared the predicted clinical score with the one given by our onsite clinician and saw a very good correlation between our prediction and what the clinician might predict,” said Gholami.
The algorithms mathematically de fined the MS subjects’ gait features at different severity levels, determining their level of gait abnormality. Principal component and linear discriminant analyses were performed on all the captured gait indices to distinguish the patient population from the control subjects, and quantify the level of progression of the MS disease in patients.
The imaging technique will allow clinical physicians to understand the effectiveness of certain medications on a patient. By finding subtle changes in gait pattern, ones not detectable necessarily by pure observation, doctors can ascertain improvements or worsening conditions.
The Future in Motion
The portable Kinect device provides an inexpensive alternative to marker-based motion-capture technologies, which are more accurate but expensive and difficult to transport.
With sophisticated motion capture technologies, infrared cameras track markers placed on specific anatomical landmarks of a subject. Based on the information of the markers, re searchers can reconstruct an individual’s motion.
“These cameras are very expensive,” said Gholami. “They are about $50,000 to $300,000 in comparison with the Microsoft Kinect, which is $200.”
Additionally, Gholami sees the imaging algorithm being used by pharmaceutical companies as the firms test the effects of new medications. The McGill researcher also sees remote monitoring and sensing capabilities.
“There is no need for the patient to visit the clinic,” said Gholami. “All the information can be captured at the place that the patient lives, and can be sent to the clinic for further assessment and evaluation.”
Gholami next hopes to refine the software algorithm and incorporate a larger population than the ten test patients. The team is also working on a prototype that can be easily used in clinics.
Gholami, supervised by Jozsef Kövecses from the Department of Mechanical Engineering and Centre for Intelligent Machines, partnered with Daria Trojan, a physiatrist in the Department of Neurology and Neurosurgery working at the Montreal Neurological Institute and Hospital.
The work was completed in collaboration with Behnood Gholami at the Hoboken, NJ-based AreteX Systems Inc. and Wassim M. Haddad from Georgia Institute of Technology.
The team’s research paper, “A Microsoft Kinect-Based Point-of-Care Gait Assessment Framework for Multiple Sclerosis Patients,” was published in the IEEE Journal of Biomedical and Health Informatics on July 21, 2016.