An Alternative Myoelectric Pattern Recognition Approach for the Control of Hand Prostheses: A Case Study of Use in Daily Life by a Dysmelia Subject

Abstract

The functionality of upper limb prostheses can be improved by intuitive control strategies that use bioelectric signals measured at the stump level. One such strategy is the decoding of motor volition via myoelectric pattern recognition (MPR), which has shown promising results in controlled environments and more recently in clinical practice. Moreover, not much has been reported about daily life implementation and real-time accuracy of these decoding algorithms. This paper introduces an alternative approach in which MPR allows intuitive control of four different grips and open/close in a multifunctional prosthetic hand. We conducted a clinical proof-of-concept in activities of daily life by constructing a self-contained, MPR-controlled, transradial prosthetic system provided with a novel user interface meant to log errors during real-time operation. The system was used for five days by a unilateral dysmelia subject whose hand had never developed, and who nevertheless learned to generate patterns of myoelectric activity, reported as intuitive, for multi-functional prosthetic control. The subject was instructed to manually log errors when they occurred via the user interface mounted on the prosthesis. This allowed the collection of information about prosthesis usage and real-time classification accuracy. The assessment of capacity for myoelectric control test was used to compare the proposed approach to the conventional prosthetic control approach, direct control. Regarding the MPR approach, the subject reported a more intuitive control when selecting the different grips, but also a higher uncertainty during proportional continuous movements. This paper represents an alternative to the conventional use of MPR, and this alternative may be particularly suitable for a certain type of amputee patients. Moreover, it represents a further validation of MPR with dysmelia cases.

Keywords: Prosthetic control; assessment of capacity for myoelectric control (ACMC); dysmelia; electromyogram (emg); myoelectric pattern recognition (MPR).

FIGURE 1.

a) Representation of the MPR-based system to control the multifunctional prosthetic hand. The prosthesis allows four different grips achievable by doing preconfigured patterns: hold open, co-contraction, a double open impulse, and a triple open impulse. These patterns encoded the grips open palm, side grip, fine grip and pointer respectively. The MPR controller (Artificial Limb Controller) worked as a translational circuit in a semi-sequential trend, operating the hand according to the predicted posture. Each grip was then operated via proportional open and close hand movements. A majority vote buffer was used to reduce misclassifications. b) Representation of the error signaling system implemented for the Continuous Monitoring test. An array of buttons was placed on the outer socket of the prosthetic. In case of perceived misclassification, pressing an error button puts the controller in a temporary stand-by state forcing it to wait for the user to indicate the desired movement which was wrongly executed. The system then saves the occurrence of the error and the intended movement.

FIGURE 2.

Movements investigated using myoelectric pattern recognition.

FIGURE 3.

Electrode placement over the transradial dysmelia subject’s residual limb, inner (left) and outer (right) sides. The reference electrode was placed on the bony part of the elbow.

FIGURE 4.

Validation of real-time myoelectric pattern recognition on the dysmelia subject. Results of the Motion Test for different channels/movements configurations (OH = open hand, CH = close hand, SG = side grip, FG = fine grip, PTR = pointer, PRO = pronation, SUP = supination, AVG = average). The symbol and the line in the boxplots represent the mean and median of each box, respectively. Accuracy was calculated using the predictions during the completion time, and only completed motions contributed. Completion Rate is the rate of successful trials. Selection Time is the time required to reach the first correct prediction. Completion Time is the time to reach 20 correct predictions.

FIGURE 5.

Amount of manually reported errors per hour over the five days of the Continuous Monitoring test.

FIGURE 6.

Average duration time and muscular effort for series of continuous proportional movements (open and close hand). The numbers above the bars represent the amount of series registered for that particular movement along that day. The strength was calculated over the mean absolute value as the averaged proportional value to the maximum registered value (averaged between all channels) over all days.

FIGURE 7.

Real-time myoelectric pattern recognition accuracy. Confusion matrices, for classification (left) and execution (right), resulted from post-analysis of the data logged during the Continuous Monitoring test. All values are presented in percentage (OH = open hand, CH = close hand, SG = side grip, FG = fine grip, PTR = pointer, RST = rest).

FIGURE 8.

Assessment of Capacity for Myoelectric Control (ACMC). The figures show the three functional tasks involved in the test: building a ready-to-assemble lamp project (a lamp), wrapping a present and writing a gift card, and setting up a table for six people.