Application of computational intelligence for automated diagnosis in gait analysis
Professor Dr. Hans Jürgen Gerner
Zentrum für Orthopädie, Unfallchirurgie und Paraplegiologie
Klinik Orthopädie und Unfallchirurgie
Professor Dr.-Ing. Georg Bretthauer
Karlsruher Institut für Technologie (KIT)
Institut für Angewandte Informatik
Behandlungszentrum Aschau GmbH
DescriptionThe Instrumental Gait Analysis is an important method for a quantitative observation and documentation of pathological human movements. Therefore, data were recorded obtained from video measurements, ground reaction forces, muscles activities (EMG, electromyography), or oxygen consumption (to determine walk energy). This procedure provides dynamic behaviors, which are very complex to observe and evaluate by humans. Although, these data were used for clinical applications to support diagnosis and therapy decisions for handicapped humans, to observe movements with respect to artificial muscles electro stimulation, to control prostheses, and to get experience for walking robots.
For this project, the instrumental gait analysis is taken into account to get diagnosis hints and to help planning a therapy. The observed patients suffer from partial paraplegia (approx. 1,200 new cases in Germany each year), Cerebral Palsy (approx. 20,000 cases p.a.), and apoplexia (approx. 150,000 cases p.a.). Thereby, sometimes only vague physician's decision rules exist, whereupon an examination is assessed. This problem retards a routine application for all patients groups, or an apprenticeship for young physicians, even if the gait laboratory's technical design and constructions are very well-engineered.
New trials were carried out to get decision-rules by a computer with respect to build up an expert knowledge. Therefore, methods were used for only specific applications. The main problem is to figure out suitable computer phrases that represents human's knowledge. However, the trust is quite poor in new computer aided diagnosis developments. The reason is most of the methods and outputs are hard to interpret or to understand.
In collaboration with the University of Heidelberg (Gait Laboratory and Treadmill locomotion), it was tried to design data-analysis methods that are based on expert knowledge and that are able to gain this knowledge. With it, the physician's decisions should be supported with a quantitative assessment. At least, the patients get more confidence with statements that rely on data-proofed cognition.
The algorithms were programmed in MATLAB, and partly built in the existing commercial gait system in Heidelberg.
Wolf, S.; Loose, T.; Schablowski, M.; Döderlein, L.; Rupp, R.; Gerner, H. J.; Bretthauer, G. & Mikut, R.: Automated Feature Assessment in Instrumented Gait Analysis. Gait & Posture, 2006, 23(3), 331-338
Wolf, S.; Braatz, F.; Metaxiotis, D.; Armbrust, P.; Dreher, T.; Döderlein, L. & Mikut, R.: Gait analysis may help to distinguish hereditary spastic paraplegia from cerebral palsy. Gait & Posture, 2011, 33(4), 556-561
Wolf, S. I.; Mikut, R.; Kranzl, A. & Dreher, T.: Which Functional Impairments are the Main Contributors to Pelvic Anterior Tilt during Gait in Individuals with Cerebral Palsy?
Gait & Posture, Elsevier, 2014, 39, 359-364