Development of biomimetic neural sensory feedback for neuroprosthetic applications

Re-establishing the sensory flow of information between prostheses and the brain is of paramount importance. Lack of sensory feedback and inadequate embodiment are among the reasons for rejection of available commercial prosthesis. Implantable peripheral nerve interfaces can be reliably used to provide sensory feedback to limb amputees. This approach can improve prosthesis control, is usable and stable over long periods, and fosters embodiment of the prosthesis by the subjects. Previous works also showed that different encoding strategies could be used to successfully restore sensory feedback. In particular, in these studies the amplitude or frequency of the injected stimuli was modulated, eliciting somatotopic sensations (e.g., referred to phantom extremity) with feelings that are sometimes similar to the natural ones (e.g., pressure or vibration). However, naturalness can be characterized by different ‘‘grades’’ and the ‘‘level’’ of naturalness has been often reported by the subjects as quite limited (quite unpleasant). Ideally, neural stimulation should be able to provide sensory feedback that is functionally effective and highly natural, as the naturalness of the feedback plays a pivotal role in prostheses acceptance. A possible way to address this issue might be to define and use complex stimulation patterns that resemble the natural encoding strategies implemented by the nervous system: the biomimetic neural stimulation strategies. This paradigm should be implemented in a real-time configuration in a bionic prosthesis.

Re-establishing the sensory flow of information between prostheses and the brain is of paramount importance. Lack of sensory feedback and inadequate embodiment are among the reasons for rejection of available commercial prosthesis. Implantable peripheral nerve interfaces can be reliably used to provide sensory feedback to limb amputees. This approach can improve prosthesis control, is usable and stable over long periods, and fosters embodiment of the prosthesis by the subjects. Previous works also showed that different encoding strategies could be used to successfully restore sensory feedback. In particular, in these studies the amplitude or frequency of the injected stimuli was modulated, eliciting somatotopic sensations (e.g., referred to phantom extremity) with feelings that are sometimes similar to the natural ones (e.g., pressure or vibration). However, naturalness can be characterized by different ‘‘grades’’ and the ‘‘level’’ of naturalness has been often reported by the subjects as quite limited (quite unpleasant). Ideally, neural stimulation should be able to provide sensory feedback that is functionally effective and highly natural, as the naturalness of the feedback plays a pivotal role in prostheses acceptance.
A possible way to address this issue might be to define and use complex stimulation patterns that resemble the natural encoding strategies implemented by the nervous system: the biomimetic neural stimulation strategies. This paradigm should be implemented in a real-time configuration in a bionic prosthesis.

The student will be guided in understanding the principal causes of lack of sensory feedback, its effects and meaning in terms of artificial perception, the current state of the art of neuroprosthesis with scientific literature readings, and our developed sensory-feedback system. The student will study in detail the mechanisms and principles of direct electrical nerve stimulation for sensory feedback applications. The major goals (mandatory) for the student will be: 1. Processing of neural signals recorded experimentally with neural electrodes implanted in the nervous systems, in response to different neural stimulation. 2. Extraction of key features of biomimicry from different neural signals. 3. Design and implementation of a novel biomimetic sensory encoding based on mechano-neural model of touch (FootSim/TouchSim). 4. Real-time implementation of a biomimetic neuromodulation according to wearable sensors outputs in a closed-loop neuroprosthetic system. 5. Validation of the new system with volunteers/amputees Recommendable skills: Signal processing, MATLAB, C++, C, peripheral nervous system neurophysiology and anatomy. Extra skills: Computational neuroscience, multithread applications, embedded linux system (e.g. raspberry pi).

The student will be guided in understanding the principal causes of lack of sensory feedback, its effects and meaning in terms of artificial perception, the current state of the art of neuroprosthesis with scientific literature readings, and our developed sensory-feedback system. The student will study in detail the mechanisms and principles of direct electrical nerve stimulation for sensory feedback applications.
The major goals (mandatory) for the student will be:
1. Processing of neural signals recorded experimentally with neural electrodes implanted in the nervous systems, in response to different neural stimulation.
2. Extraction of key features of biomimicry from different neural signals.
3. Design and implementation of a novel biomimetic sensory encoding based on mechano-neural model of touch (FootSim/TouchSim).
4. Real-time implementation of a biomimetic neuromodulation according to wearable sensors outputs in a closed-loop neuroprosthetic system.
5. Validation of the new system with volunteers/amputees
Recommendable skills: Signal processing, MATLAB, C++, C, peripheral nervous system neurophysiology and anatomy.
Extra skills: Computational neuroscience, multithread applications, embedded linux system (e.g. raspberry pi).

Dr. Giacomo Valle, Assistant Professor, Head of Neural Bionics laboratory, Chalmers University of Technology, Life Bionics, Goteborg Sweden

Dr. Giacomo Valle, Assistant Professor, Head of Neural Bionics laboratory, Chalmers University of Technology, Life Bionics, Goteborg Sweden