Patent Application
20130184539
The invention relates to a combined sensor system for deriving muscle signals on the skin surface to control, for example, hand or leg prostheses, or as an input for computers or other devices.
It is known to arrange myoelectric sensors close to muscles on the skin surface, the sensors detecting the depolarisation signal emitted to all sides and providing it for further processing [LUCA, ALTER, HERR09]. It is furthermore known that the depolarisation signals are overlaid by simultaneously stimulated, innervated muscles. This effect increases with a growing spacing of the derivation electrodes and is more pronounced in myoelectric sensors with single-differential electrodes than in sensors with double-differential electrodes [FARINA, DELSYS].
It is further indicated that near-infrared spectroscopy (NIRS), an active measuring method for the relative measurement of the blood oxygen content with at least one light transmitter and one receiver [ASLIN, CHALM], is also suitable for investigating muscle activity {AKIMA, QUARE, MIURA01 MIURA03]. The combination of a single-differential myoelectric sensor with a simple near-infrared sensor, consisting of a light source and a photo receiver to detect muscle activities is also found in the literature [HERR10A, HERR10B, HERR11A, HERR11 B]. In the same literature, it is also shown how the derived signals of the two sensors can be synchronously combined as a system by a suitable weighting. A detailed section about the weighted combination of two features in one time window was disclosed in [HERR10C].
Using the previously known combinations of a myoelectric and near-infrared sensor it is only possible to detect one region under the sensor, which cannot be varied during the measurement. The overlays of the myoelectric signals of a plurality of muscles are recorded at a recording point and allow the basic distinction between activity and the rest state and the distinction between different contraction types. However, there is no information about the spatial origin of the individual signal components. A near-infrared sensor, in contrast to this, only detects a spatially narrowly limited region below the sensor. This sensor type only indirectly allows an inference of muscle activity, as the blood quantity, and therefore the quantity of the haemoglobin, is influenced by a contraction.
The problem to be solved is therefore that a control signal proportional to the muscle contraction has to be derived, allowing a statement about a spatially narrowly limited, well definable region in the musculature with minimal interferences and without signal artefacts. The near-infrared signal with a high spatial resolution cannot be successfully used on its own to control a technical arrangement or, for example, a prosthesis. The proportionality between the muscle force exerted and the measured signal amplitude is lacking for this.
The problem disclosed is solved in that at least one double-differential myoelectric sensor and at least one near-infrared light transmitter and receiver are arranged on the skin surface, in each case, in order to detect both the muscle activity and the blood oxygen content, or the relative change in the quantity of haemoglobin, of the muscle being observed. If the blood oxygen content or the quantity of haemoglobin in the observation region changes, less light reaches the near-infrared receiver from the near-infrared transmitter. The blood oxygen content can be determined by means of the light intensity, as explained in the prior art and the contraction inferred. The penetration depth of the light into the tissue, and therefore the observation depth, is determined by the spacing apart of the NIR transmitter and NIR receiver. As different movements have characteristic time-variant patterns in the EMG and NIR signal, the movement can be inferred by measuring the two signals and thus, for example, an arm prosthesis can be controlled successfully and in a manner close to reality.
The control signals for an arm, hand or leg prosthesis, for example, are produced by fusing the sensor signals and by combining the sensors. The myoelectric signal is used here for the general recognition of the beginning of a contraction. Following this in terms of time, in a short time period, the muscle contraction is classified within a spatially limited region by means of an NIR signal. When using a plurality of NIR sensors, the reliability of the classification and moreover also muscle activities in different muscle regions are distinguished and classified. Following the classification phase, which is short in terms of time, a movement or action of the technical arrangement, for example a prosthesis, can be carried out depending on the myoelectric signal(s).
The invention consists of a plurality of electrodes and one or more near-infrared light sources and one or more optical receivers, which are arranged in a spatial relationship to one another.
The embodiment in
Two or more centrally arranged EMG sensors (13) and (14) allow the myoelectric signals of a large muscle, for example on the leg, or a plurality of muscles located next to one another, for example the chest region, to be detected simultaneously. The signals of the sensors (13) and (14) are used here either individually or as correlated signals, for example to reduce the noise or remove interferences or artefacts. One embodiment is shown in
In all the arrangements, the myoelectric and the near-infrared signal are not evaluated simultaneously, but consecutively. The reason for this is the different signal properties and the information contained therein, which are based on various underlying physical effects. The mean value of myoelectric signal is approximately proportional to the muscle force exerted. However, inferences about the type of contraction exerted can hardly be drawn from a single myoelectric signal, as an overlay of signals of a plurality of muscles occurs, it not being possible to determine the origin. If, as, for example, in the forearm, a plurality of muscles are located in the surroundings of the EMG sensor, a distinction cannot be made as to which one has carried out a contraction. Owing to the fixed and spatially delimited observation region of the near-infrared sensor, different contractions can be inferred with the aid of a sensor. However, the resulting signal is not proportional to the force developed and also only supplies meaningful information at the beginning of a contraction. For this reason, the myoelectric signal is firstly used for recognition as to whether a muscle contraction is present at all. The NIR signal cannot be used for this, as it is subject to interferences due to pressure changes or changes in the surrounding light. Once a contraction has been recognised, a short time period can be used in the NIR signal to classify the type of contraction. As no inferences as to the strength of the contraction are possible by this, however, the myoelectric signal is used again following the classification in order to control a movement proportionally, in other words depending on the muscle force exerted.
Multiple NIRS imaging device to the exercising muscle metabolism. Spectroscopy, Vol. 17, pp. 549-558.
| Number | Date | Country | Kind |
|---|---|---|---|
| 102011118732.8 | Nov 2011 | DE | national |
