![Illustration of a bidirectional prosthetic hand with sensory feedback and EMG pattern recognition. The tactile information collected by sensors (a) is delivered to the central nervous system (c) through sensory information coding (b) along the sensory afferent pathway. The motor intentions are identified by pattern recognition from iEMG/sEMG (such as HD-sEMG) signals (d) to control prosthetic hands (e). This figure is modified from [2] with permission. Reprinted from [2], Copyright (2020), with permission from Elsevier.](https://www.researchgate.net/publication/369420181/figure/fig1/AS:11431281307619652@1738702671416/Illustration-of-a-bidirectional-prosthetic-hand-with-sensory-feedback-and-EMG-pattern_Q320.jpg)
![Sensory encoding and afferent pathway of ETS. TENS is applied on the PFM area (a) and (c) via surface electrodes (Reproduced from [105]. © IOP Publishing Ltd. CC BY 3.0.). (b) Encoding scheme for multiple sensory information. (d) Electrodes that activate the sensory receptors and nerves beneath the skin (e). (f) The afferent pathway way and (g) active regions in primary somatosensory cortex (SI) at the amputated side are symmetric to that at the contralateral hand. Reproduced from [24]. CC BY 4.0.](https://www.researchgate.net/publication/369420181/figure/fig2/AS:11431281307649327@1738702671576/Sensory-encoding-and-afferent-pathway-of-ETS-TENS-is-applied-on-the-PFM-area-a-and-c_Q320.jpg)
![Prototype of single flexor control of a cable-driven prosthetic hand with a biorealistic muscle reflex model. Residual sEMG from the individual with amputation passes the Bayesian filter to obtain the alpha motor command to establish a model-calculated force to drive the torque motor (e). Torque from the motor pulls a cable to create prosthetic hand movements (f). The biorealistic muscle reflex model contains 768 spiking neurons (a), a skeletal muscle (b), and a spindle (c) projecting 128 Ia afferents, which is all implemented in the neuromorphic chip (d) for real-time simulation. Length–tension and force–velocity properties are incorporated in the muscle reflex model. In this biorealistic reflex loop, the alpha motor command enters the model as an excitatory postsynaptic current, which is delivered to the motoneuron pool (a). Its output spike signals are subsequently converted to excitatory input that activates the Hill-type model of skeletal muscle (b) after a twitch model. The muscle model (b) calculates a muscle force output that drives the torque motor. Information about muscle lengthening is sent back to a model of muscle spindle (c), which subsequently produces excitatory afferents back to the motoneuron pool (a). This figure is adapted from [25] with permission. Reproduced from [25]. CC BY 4.0.](https://www.researchgate.net/publication/369420181/figure/fig3/AS:11431281307600242@1738702671727/Prototype-of-single-flexor-control-of-a-cable-driven-prosthetic-hand-with-a-biorealistic_Q320.jpg)
April 2023
·
94 Reads
·
8 Citations
Significant advances have been made to improve control and to provide sensory functions for bionic hands. However, great challenges remain, limiting wide acceptance of bionic hands due to inadequate bidirectional neural compatibility with human users. Recent research has brought to light the necessity for matching neuromechanical behaviors between the prosthesis and the sensorimotor system of amputees. A novel approach to achieving greater neural compatibility leverages the technology of biorealistic modeling with real-time computation. These studies have demonstrated a promising outlook that this unique approach may transform the performance of hand prostheses. Simultaneously, a noninvasive technique of somatotopic sensory feedback has been developed based on evoked tactile sensation (ETS) for conveying natural, intuitive, and digit-specific tactile information to users. This paper reports the recent work on these two important aspects of sensorimotor functions in prosthetic research. A background review is presented first on the state of the art of bionic hand and the various techniques to deliver tactile sensory information to users. Progress in developing the novel biorealistic hand prosthesis and the technique of noninvasive ETS feedback is then highlighted. Finally, challenges to future development of the biorealistic hand prosthesis and implementing the ETS feedback are discussed with respect to shaping a next-generation hand prosthesis.
![Illustration of neural control of a human hand (a) and biomimetic control of a prosthetic hand (b). Before amputation, the brain controls human hand through an intact neuromuscular reflex system along with proprioceptive and cutaneous sensory afferents. In prosthetic control, the neuromuscular reflex process is amiss, and the sensory feedback information provided may be limited and incompatible to what is acquainted to the brain. Thus, it is essential to restore the neuromuscular reflex process and natural sensory feedback for prosthetic hand. Visual information determines hand positioning and opening. Abbreviations: HD-EMG: high-density EMG; ETS: evoked tactile sensation. Part (a) is modified from Figure 2 in [10] with permission.](https://www.researchgate.net/publication/350402431/figure/fig1/AS:11431281222593953@1707303373751/Illustration-of-neural-control-of-a-human-hand-a-and-biomimetic-control-of-a-prosthetic_Q320.jpg)
