Development of the human interface equipment based on surface EMG employing channel selection method

Basic study on combined motion estimation using multichannel surface EMG signals

To improve degree of freedom (DOF) of system control using surface electromyogram (SEMG), we made a basic study of the estimation of user's intended motion including combined motion which is performed by more than one basic motion simultaneously. Our developed system requested to obtain three SEMG characteristics of basic motion and one SEMG characteristics of rest state. This study defines the motion of grasp, supination and pronation as basic motion, and two combined motion which is "grasp +supination" and "grasp + pronation" are set. Our system investigates the possibility of combined motion estimation based on SEMG characteristics of basic motions. Estimation method which is utilizing optimal SEMG that are derived from multichannel SEMG signals is performed by canonical discriminant space and tendency of degree of similarity between combined motion and basic motion. In experimental results, we succeeded in estimation of combined motion although it was included an estimation of basic motions which were constructed elements of combined motion.

Real-time decision fusion for multimodal neural prosthetic devices

BACKGROUND

The field of neural prosthetics aims to develop prosthetic limbs with a brain-computer interface (BCI) through which neural activity is decoded into movements. A natural extension of current research is the incorporation of neural activity from multiple modalities to more accurately estimate the user's intent. The challenge remains how to appropriately combine this information in real-time for a neural prosthetic device.

METHODOLOGY/PRINCIPAL FINDINGS

Here we propose a framework based on decision fusion, i.e., fusing predictions from several single-modality decoders to produce a more accurate device state estimate. We examine two algorithms for continuous variable decision fusion: the Kalman filter and artificial neural networks (ANNs). Using simulated cortical neural spike signals, we implemented several successful individual neural decoding algorithms, and tested the capabilities of each fusion method in the context of decoding 2-dimensional endpoint trajectories of a neural prosthetic arm. Extensively testing these methods on random trajectories, we find that on average both the Kalman filter and ANNs successfully fuse the individual decoder estimates to produce more accurate predictions.

CONCLUSIONS

Our results reveal that a fusion-based approach has the potential to improve prediction accuracy over individual decoders of varying quality, and we hope that this work will encourage multimodal neural prosthetics experiments in the future.

Estimation of muscle strength during motion recognition using multichannel surface EMG signals

The use of kinesiological electromyography is established as an evaluation tool for various kinds of applied research, and surface electromyogram (SEMG) has been widely used as a control source for human interfaces such as in a myoelectric prosthetic hand (we call them 'SEMG interfaces'). It is desirable to be able to control the SEMG interfaces with the same feeling as body movement. The existing SEMG interface mainly focuses on how to achieve accurate recognition of the intended movement. However, detecting muscular strength and reduced number of electrodes are also an important factor in controlling them. Therefore, our objective in this study is the development of and the estimation method for muscular strength that maintains the accuracy of hand motion recognition to reflect the result of measured power in a controlled object. Although the muscular strength can be evaluated by various methods, in this study a grasp force index was applied to evaluate the muscular strength. In order to achieve our objective, we directed our attention to measuring all valuable information for SEMG. This work proposes an application method of two simple linear models, and the selection method of an optimal electrode configuration to use them effectively. Our system required four SEMG measurement electrodes in which locations differed for every subject depending on the individual's characteristics, and those were selected from a 96ch multi electrode using the Monte Carlo method. From the experimental results, the performance in six normal subjects indicated that the recognition rate of four motions were perfect and the grasp force estimated result fit well with the actual measurement result.