Modelo neuromórfico para estimulação transcutânea e otimização dos níveis de percepção sensorial

Abstract

It is estimated that in Brazil about 16,064 people have suffered some kind of amputation or disarticulation of the upper limb in the last 10 years. To restore parts of motor and sensory functionalities lost with amputation it is necessary to use prostheses, however, even the most advanced commercial prostheses, such as myoelectrics, are not able to restore important functionalities, resulting in a high rate of abandonment of the prosthetic limb. Studies indicate that one of the main reasons for abandonment is the absence of tactile sensory feedback, which is responsible for estimating the texture, shape, and complacency of different materials, allowing the user to adjust the forces required in the manipulation of objects, thus avoiding slipping. In the absence of tactile feedback, users become dependent on other types of feedback, such as visual or auditory, which are not always effective. Thus, it is necessary to develop strategies capable of providing informative tactile feedback to those users. Surface electrostimulation can be a viable technique in restoring this functionality, since it is capable of provide different modalities of sensation, such as pressure and vibration. However, it is necessary to develop a model capable to combine the output of high density artificial tactile sensors into a few electro-stimulation channels and determine the best modulation of the stimulation parameters to provide discriminable electrotactile stimuli. Aiming to overcome this limitation, this work proposes a neuromorphic model, inspired by the intracellular dynamics of the cuneiform nucleus, capable of coding and optimizing information coming from high-density artificial tactile sensors, which can be used in the future to restore the tactile feedback in amputees. This model was implemented combining signals from 16 taxels, modeled as primary afferents of type 1, into a single spike output. In sequence, this study seeks to characterize the sensation caused by transcutaneous electro-stimulation modulated in frequency and amplitude, to enable the optimization of the sensory perception of the stimuli generated by the bioinspired model. The results of this work demonstrate that the proposed neuromorphic model can provide a distinct triggering output for different types of stimuli, here the capacity of this model to separate eight naturalistic textures was evaluated, reinforcing the immense capacity of segregation and generalization of information of the cuneiform neuron modeled. Evidencing that this model can reestablish the adequate restoration of tactile feedback that would even allow the user to distinguish different groups of naturalistic textures with only one channel of electro-stimulation, provided that the sensory perception of the end-user is considered. Since it has been demonstrated that the perception of electro-stimulation varies individually, and that the amplitude of stimulation directly influences the perception of frequency variation. It is necessary to determine individually, depending on the stimulation site and pulse width, the stimulation amplitude, and frequency range, for the proper optimization of the sensorial perceptions and modulation of the output of the bioinspired model proposed here for the adequate restoration of the tactile sensory feedback.