Electrocardiogram Delineation Using Deep Neural Networks.

BACKGROUND: In recent years, there has been a rising interest in the application of deep neural networks (DNN) for the delineation of the electrocardiogram (ECG). OBJECTIVES: A variety of DNN architectures has been investigated in a 5-fold cross-validation approach. RESULTS: The best performing network achieved 100% sensitivity and >97% positive predictive value for all ECG waves. CONCLUSION: Our DNN could achieve similar classification performance as other DNN approaches described in the literature at a reduced computational cost.

Increased Symmetry of Lower-Limb Amputees Walking With Concurrent Bilateral Vibrotactile Feedback.

Gait asymmetry in lower-limb amputees can lead to several secondary conditions that can decrease general health and quality of life. Including augmented sensory feedback in rehabilitation programs can effectively mitigate spatiotemporal gait irregularities. Such benefits can be obtained with non-invasive haptic systems representing an advantageous choice for usability in overground training and every-day life. In this study, we tested a wearable tactile feedback device delivering short-lasting (100ms) vibrations around the waist syncronized to gait events, to improve the temporal gait symmetry of lower-limb amputees. Three above-knee amputees participated in the study. The device provided bilateral stimulations during a training program that involved ground-level gait training. After three training sessions, participants showed higher temporal symmetry when walking with the haptic feedback in comparison to their natural walking (resulting symmetry index increases of +2.8% for Subject IDA, +12.7% for Subject IDB and +2.9% for Subject IDC). One subject retained improved symmetry (Subject IDB,+14.9%) even when walking without the device. Gait analyses revealed that higher temporal symmetry may lead to concurrent compensation strategies in the trunk and pelvis. Overall, the results of this pilot study confirm the potential utility of sensory feedback devices to positively influence gait parameters when used in supervised settings. Future studies shall clarify more precisely the training modalities and the targets of rehabilitation programs with such devices.

A Multi-Parametric Wearable System to Monitor Neck Movements and Respiratory Frequency of Computer Workers.

Musculoskeletal disorders are the most common form of occupational ill-health. Neck pain is one of the most prevalent musculoskeletal disorders experienced by computer workers. Wrong postural habits and non-compliance of the workstation to ergonomics guidelines are the leading causes of neck pain. These factors may also alter respiratory functions. Health and safety interventions can reduce neck pain and, more generally, the symptoms of musculoskeletal disorders and reduce the consequent economic burden. In this work, a multi-parametric wearable system based on two fiber Bragg grating sensors is proposed for monitoring neck movements and breathing activity of computer workers. The sensing elements were positioned on the neck, in the frontal and sagittal planes, to monitor: (i) flexion-extension and axial rotation repetitions, and (ii) respiratory frequency. In this pilot study, five volunteers were enrolled and performed five repetitions of both flexion-extension and axial rotation, and ten breaths of both quite breathing and tachypnea. Results showed the good performances of the proposed system in monitoring the aforementioned parameters when compared to optical reference systems. The wearable system is able to well-match the trend in time of the neck movements (both flexion-extension and axial rotation) and to estimate mean and breath-by-breath respiratory frequency values with percentage errors ≤6.09% and ≤1.90%, during quiet breathing and tachypnea, respectively.

Morphological Neural Computation Restores Discrimination of Naturalistic Textures in Trans-radial Amputees.

Humans rely on their sense of touch to interact with the environment. Thus, restoring lost tactile sensory capabilities in amputees would advance their quality of life. In particular, texture discrimination is an important component for the interaction with the environment, but its restoration in amputees has been so far limited to simplified gratings. Here we show that naturalistic textures can be discriminated by trans-radial amputees using intraneural peripheral stimulation and tactile sensors located close to the outer layer of the artificial skin. These sensors exploit the morphological neural computation (MNC) approach, i.e., the embodiment of neural computational functions into the physical structure of the device, encoding normal and shear stress to guarantee a faithful neural temporal representation of stimulus spatial structure. Two trans-radial amputees successfully discriminated naturalistic textures via the MNC-based tactile feedback. The results also allowed to shed light on the relevance of spike temporal encoding in the mechanisms used to discriminate naturalistic textures. Our findings pave the way to the development of more natural bionic limbs.

Cuneate spiking neural network learning to classify naturalistic texture stimuli under varying sensing conditions.

We implemented a functional neuronal network that was able to learn and discriminate haptic features from biomimetic tactile sensor inputs using a two-layer spiking neuron model and homeostatic synaptic learning mechanism. The first order neuron model was used to emulate biological tactile afferents and the second order neuron model was used to emulate biological cuneate neurons. We have evaluated 10 naturalistic textures using a passive touch protocol, under varying sensing conditions. Tactile sensor data acquired with five textures under five sensing conditions were used for a synaptic learning process, to tune the synaptic weights between tactile afferents and cuneate neurons. Using post-learning synaptic weights, we evaluated the individual and population cuneate neuron responses by decoding across 10 stimuli, under varying sensing conditions. This resulted in a high decoding performance. We further validated the decoding performance across stimuli, irrespective of sensing velocities using a set of 25 cuneate neuron responses. This resulted in a median decoding performance of 96% across the set of cuneate neurons. Being able to learn and perform generalized discrimination across tactile stimuli, makes this functional spiking tactile system effective and suitable for further robotic applications.

Tactile Sensing and Control of Robotic Manipulator Integrating Fiber Bragg Grating Strain-Sensor.

Tactile sensing is an instrumental modality of robotic manipulation, as it provides information that is not accessible via remote sensors such as cameras or lidars. Touch is particularly crucial in unstructured environments, where the robot's internal representation of manipulated objects is uncertain. In this study we present the sensorization of an existing artificial hand, with the aim to achieve fine control of robotic limbs and perception of object's physical properties. Tactile feedback is conveyed by means of a soft sensor integrated at the fingertip of a robotic hand. The sensor consists of an optical fiber, housing Fiber Bragg Gratings (FBGs) transducers, embedded into a soft polymeric material integrated on a rigid hand. Through several tasks involving grasps of different objects in various conditions, the ability of the system to acquire information is assessed. Results show that a classifier based on the sensor outputs of the robotic hand is capable of accurately detecting both size and rigidity of the operated objects (99.36 and 100% accuracy, respectively). Furthermore, the outputs provide evidence of the ability to grab fragile objects without breakage or slippage e and to perform dynamic manipulative tasks, that involve the adaptation of fingers position based on the grasped objects' condition.

Intracellular Dynamics in Cuneate Nucleus Neurons Support Self-Stabilizing Learning of Generalizable Tactile Representations.

How the brain represents the external world is an unresolved issue for neuroscience, which could provide fundamental insights into brain circuitry operation and solutions for artificial intelligence and robotics. The neurons of the cuneate nucleus form the first interface for the sense of touch in the brain. They were previously shown to have a highly skewed synaptic weight distribution for tactile primary afferent inputs, suggesting that their connectivity is strongly shaped by learning. Here we first characterized the intracellular dynamics and inhibitory synaptic inputs of cuneate neurons in vivo and modeled their integration of tactile sensory inputs. We then replaced the tactile inputs with input from a sensorized bionic fingertip and modeled the learning-induced representations that emerged from varied sensory experiences. The model reproduced both the intrinsic membrane dynamics and the synaptic weight distributions observed in cuneate neurons in vivo. In terms of higher level model properties, individual cuneate neurons learnt to identify specific sets of correlated sensors, which at the population level resulted in a decomposition of the sensor space into its recurring high-dimensional components. Such vector components could be applied to identify both past and novel sensory experiences and likely correspond to the fundamental haptic input features these neurons encode in vivo. In addition, we show that the cuneate learning architecture is robust to a wide range of intrinsic parameter settings due to the neuronal intrinsic dynamics. Therefore, the architecture is a potentially generic solution for forming versatile representations of the external world in different sensor systems.

Bilateral Tactile Input Patterns Decoded at Comparable Levels But Different Time Scales in Neocortical Neurons.

The presence of contralateral tactile input can profoundly affect ipsilateral tactile perception, and unilateral stroke in somatosensory areas can result in bilateral tactile deficits, suggesting that bilateral tactile integration is an important part of brain function. Although previous studies have shown that bilateral tactile inputs exist and that there are neural interactions between inputs from the two sides, no previous study explored to what extent the local neuronal circuitry processing contains detailed information about the nature of the tactile input from the two sides. To address this question, we used a recently introduced approach to deliver a set of electrical, reproducible, tactile afferent, spatiotemporal activation patterns, which permits a high-resolution analysis of the neuronal decoding capacity, to the skin of the second forepaw digits of the anesthetized male rat. Surprisingly, we found that individual neurons of the primary somatosensory can decode contralateral and ipsilateral input patterns to comparable extents. Although the contralateral input was stronger and more rapidly decoded, given sufficient poststimulus processing time, ipsilateral decoding levels essentially caught up to contralateral levels. Moreover, there was a weak but significant correlation for neurons with high decoding performance for contralateral tactile input to also perform well on decoding ipsilateral input. Our findings shed new light on the brain mechanisms underlying bimanual haptic integration.SIGNIFICANCE STATEMENT Here we demonstrate that the spiking activity of single neocortical neurons in the somatosensory cortex of the rat can be used to decode patterned tactile stimuli delivered to the distal ventral skin of the second forepaw digits on both sides of the body. Even though comparable levels of decoding of the tactile input were achieved faster for contralateral input, given sufficient integration time each neuron was found to decode ipsilateral input with a comparable level of accuracy. Given that the neocortical neurons could decode ipsilateral inputs with such small differences between the patterns suggests that S1 cortex has access to very precise information about ipsilateral events. The findings shed new light on possible network mechanisms underlying bimanual haptic processing.

Neuromorphic Vibrotactile Stimulation of Fingertips for Encoding Object Stiffness in Telepresence Sensory Substitution and Augmentation Applications.

We present a tactile telepresence system for real-time transmission of information about object stiffness to the human fingertips. Experimental tests were performed across two laboratories (Italy and Ireland). In the Italian laboratory, a mechatronic sensing platform indented different rubber samples. Information about rubber stiffness was converted into on-off events using a neuronal spiking model and sent to a vibrotactile glove in the Irish laboratory. Participants discriminated the variation of the stiffness of stimuli according to a two-alternative forced choice protocol. Stiffness discrimination was based on the variation of the temporal pattern of spikes generated during the indentation of the rubber samples. The results suggest that vibrotactile stimulation can effectively simulate surface stiffness when using neuronal spiking models to trigger vibrations in the haptic interface. Specifically, fractional variations of stiffness down to 0.67 were significantly discriminated with the developed neuromorphic haptic interface. This is a performance comparable, though slightly worse, to the threshold obtained in a benchmark experiment evaluating the same set of stimuli naturally with the own hand. Our paper presents a bioinspired method for delivering sensory feedback about object properties to human skin based on contingency-mimetic neuronal models, and can be useful for the design of high performance haptic devices.

Slippage Detection with Piezoresistive Tactile Sensors.

One of the crucial actions to be performed during a grasping task is to avoid slippage. The human hand can rapidly correct applied forces and prevent a grasped object from falling, thanks to its advanced tactile sensing. The same capability is hard to reproduce in artificial systems, such as robotic or prosthetic hands, where sensory motor coordination for force and slippage control is very limited. In this paper, a novel algorithm for slippage detection is presented. Based on fast, easy-to-perform processing, the proposed algorithm generates an ON/OFF signal relating to the presence/absence of slippage. The method can be applied either on the raw output of a force sensor or to its calibrated force signal, and yields comparable results if applied to both normal or tangential components. A biomimetic fingertip that integrates piezoresistive MEMS sensors was employed for evaluating the method performance. Each sensor had four units, thus providing 16 mono-axial signals for the analysis. A mechatronic platform was used to produce relative movement between the finger and the test surfaces (tactile stimuli). Three surfaces with submillimetric periods were adopted for the method evaluation, and 10 experimental trials were performed per each surface. Results are illustrated in terms of slippage events detection and of latency between the slippage itself and its onset.

Spatiotemporal Dynamics of the Cortical Responses Induced by a Prolonged Tactile Stimulation of the Human Fingertips.

The sense of touch is fundamental for daily behavior. The aim of this work is to understand the neural network responsible for touch processing during a prolonged tactile stimulation, delivered by means of a mechatronic platform by passively sliding a ridged surface under the subject's fingertip while recording the electroencephalogram (EEG). We then analyzed: (i) the temporal features of the Somatosensory Evoked Potentials and their topographical distribution bilaterally across the cortex; (ii) the associated temporal modulation of the EEG frequency bands. Long-latency SEP were identified with the following physiological sequence P100-N140-P240. P100 and N140 were bilateral potentials with higher amplitude in the contralateral hemisphere and with delayed latency in the ipsilateral side. Moreover, we found a late potential elicited around 200 ms after the stimulation was stopped, which likely encoded the end of tactile input. The analysis of cortical oscillations indicated an initial increase in the power of theta band (4-7 Hz) for 500 ms after the stimulus onset followed a decrease in the power of the alpha band (8-15 Hz) that lasted for the remainder of stimulation. This decrease was prominent in the somatosensory cortex and equally distributed in both contralateral and ipsilateral hemispheres. This study shows that prolonged stimulation of the human fingertip engages the cortex in widespread bilateral processing of tactile information, with different modulations of the theta and alpha bands across time.

Artificial spatiotemporal touch inputs reveal complementary decoding in neocortical neurons.

Investigations of the mechanisms of touch perception and decoding has been hampered by difficulties in achieving invariant patterns of skin sensor activation. To obtain reproducible spatiotemporal patterns of activation of sensory afferents, we used an artificial fingertip equipped with an array of neuromorphic sensors. The artificial fingertip was used to transduce real-world haptic stimuli into spatiotemporal patterns of spikes. These spike patterns were delivered to the skin afferents of the second digit of rats via an array of stimulation electrodes. Combined with low-noise intra- and extracellular recordings from neocortical neurons in vivo, this approach provided a previously inaccessible high resolution analysis of the representation of tactile information in the neocortical neuronal circuitry. The results indicate high information content in individual neurons and reveal multiple novel neuronal tactile coding features such as heterogeneous and complementary spatiotemporal input selectivity also between neighboring neurons. Such neuronal heterogeneity and complementariness can potentially support a very high decoding capacity in a limited population of neurons. Our results also indicate a potential neuroprosthetic approach to communicate with the brain at a very high resolution and provide a potential novel solution for evaluating the degree or state of neurological disease in animal models.

Encapsulation of Piezoelectric Transducers for Sensory Augmentation and Substitution with Wearable Haptic Devices.

The integration of polymeric actuators in haptic displays is widespread nowadays, especially in virtual reality and rehabilitation applications. However, we are still far from optimizing the transducer ability in conveying sensory information. Here, we present a vibrotactile actuator characterized by a piezoelectric disk embedded in a polydimethylsiloxane (PDMS) shell. An original encapsulation technique was performed to provide the stiff active element with a compliant cover as an interface towards the soft human skin. The interface stiffness, together with the new geometry, generated an effective transmission of vibrotactile stimulation and made the encapsulated transducer a performant component for the development of wearable tactile displays. The mechanical behavior of the developed transducer was numerically modeled as a function of the driving voltage and frequency, and the exerted normal forces were experimentally measured with a load cell. The actuator was then tested for the integration in a haptic glove in single-finger and bi-finger condition, in a 2-AFC tactile stimulus recognition test. Psychophysical results across all the tested sensory conditions confirmed that the developed integrated haptic system was effective in delivering vibrotactile information when the frequency applied to the skin is within the 200⁻700 Hz range and the stimulus variation is larger than 100 Hz.

Vision-Based Pose Estimation for Robot-Mediated Hand Telerehabilitation.

Vision-based Pose Estimation (VPE) represents a non-invasive solution to allow a smooth and natural interaction between a human user and a robotic system, without requiring complex calibration procedures. Moreover, VPE interfaces are gaining momentum as they are highly intuitive, such that they can be used from untrained personnel (e.g., a generic caregiver) even in delicate tasks as rehabilitation exercises. In this paper, we present a novel master-slave setup for hand telerehabilitation with an intuitive and simple interface for remote control of a wearable hand exoskeleton, named HX. While performing rehabilitative exercises, the master unit evaluates the 3D position of a human operator's hand joints in real-time using only a RGB-D camera, and commands remotely the slave exoskeleton. Within the slave unit, the exoskeleton replicates hand movements and an external grip sensor records interaction forces, that are fed back to the operator-therapist, allowing a direct real-time assessment of the rehabilitative task. Experimental data collected with an operator and six volunteers are provided to show the feasibility of the proposed system and its performances. The results demonstrate that, leveraging on our system, the operator was able to directly control volunteers' hands movements.

Learning tactile skills through curious exploration.

We present curiosity-driven, autonomous acquisition of tactile exploratory skills on a biomimetic robot finger equipped with an array of microelectromechanical touch sensors. Instead of building tailored algorithms for solving a specific tactile task, we employ a more general curiosity-driven reinforcement learning approach that autonomously learns a set of motor skills in absence of an explicit teacher signal. In this approach, the acquisition of skills is driven by the information content of the sensory input signals relative to a learner that aims at representing sensory inputs using fewer and fewer computational resources. We show that, from initially random exploration of its environment, the robotic system autonomously develops a small set of basic motor skills that lead to different kinds of tactile input. Next, the system learns how to exploit the learned motor skills to solve supervised texture classification tasks. Our approach demonstrates the feasibility of autonomous acquisition of tactile skills on physical robotic platforms through curiosity-driven reinforcement learning, overcomes typical difficulties of engineered solutions for active tactile exploration and underactuated control, and provides a basis for studying developmental learning through intrinsic motivation in robots.