Wavelet applications to kinematic and EMG signals in gait analysis

Date of Award




Degree Name

Doctor of Philosophy (Ph.D.)

First Committee Member

Shihab S. Asfour, Committee Chair


The main objectives of the research was to (1) demonstrate the superiority of the wavelet transform over conventional signal processing techniques in processing kinematic and EMG signals, (2) determining a way to assess physiological behavior of neuromuscular transmissions, (3) providing accurate three dimensional angular description of the lower extremity behavior during human gait, and (4) providing stable and accurate frequency indices of the muscular behavior.In order to achieve the first objective, the wavelet transform was first introduced as an acceptable tool in dealing with typical problems of known solutions or typical data reported by previous researchers. The results showed that the asymmetrical Daubechies mother wavelet of the fourth order (Db4) at the second decomposition level was the most appropriate mother wavelet to be used to de-noise simple or complex kinematic activities. In addition, results indicated the superiority of the wavelet compression technique over conventional filtering techniques.In order to achieve the second objective, an experiment that aimed at studying the biceps brachii muscular behavior during isometric contractions was conducted. The EMG data collected was transformed using the continuous wavelet transform. The results provided, for the first time, an explanation of the muscular behavior in terms of separating slow- and fast-fiber recruitment levels.In order to achieve the third objective, a three dimensional dynamic lower extremity gait model was developed. The model allowed the homogenous transformation (typically used in robotics research) of the three dimensional Cartesian displacement raw data, measured by the "Selspot" motion analysis system relative to a fixed laboratory frame, to angular displacements measured relative to moving coordinate frames.In order to achieve the fourth objective, the kinematic and muscular responses of the lower extremity of eight subjects during normal gait were monitored. The kinematic signals recorded from the hip, knee and ankle joints of the right leg of each subject were smoothed using the Db4 discrete wavelet transform at the second decomposition level. Results indicated the accuracy of such filtering method in smoothing complex activities.The synchronized EMG signals recorded from the tibialis anterior, gastrocnemius, quadriceps and hamstrings of the right leg of each subject were processed using the continuous wavelet transform. Results indicated that the wavelet representation was able to: (1) differentiate between muscular behaviors during initial and steady-state gait cycles, (2) produce smoothed filtered frequency responses of the muscular behavior, (3) confirm the general idea of modeling the brain as a neural network that has the ability to learn the exact muscle power and fiber types needed to achieve a stable smooth muscular performance as gait progresses, and (4) suggest a cut-off frequency in the vicinity of 40 Hz that could be used to separate slow- and fast-fibers. This frequency matches the one used by the brain in order to alter the status of any of its peripherals.


Engineering, Biomedical; Engineering, Electronics and Electrical; Engineering, Mechanical

Link to Full Text