Lower limb sensory feedback restoration for terrain recognition via TENS

Lower limb sensory feedback restoration for terrain recognition via TENS 1st Andrea Demofonti CREO Lab Universit` a Campus Bio-Medico di Roma Rome, Italy [email protected] 2nd Annamaria Galluccio CREO Lab Universit` a Campus Bio-Medico di Roma Rome, Italy [email protected] 3rd Francesca Cordella CREO Lab Universit` a Campus Bio-Medico di Roma Rome, Italy [email protected] 4th Loredana Zollo CREO Lab Universit` a Campus Bio-Medico di Roma Rome, Italy [email protected] Abstract—Transcutaneous Electrical Nerve Stimulation (TENS) represent a promising approach to provide sensory feedback in lower limb prostheses, since it allows the elicitation of homologous and somatotopic sensation in a non invasive way through superficial electrodes that stimulate the underlying nerves. Since ground unevenness is one of the most influential environmental factors affecting the risk of falls in amputees, restoring the ability to discriminate floor conditions could play a crucial role in improving prosthesis usability and safety. However, the potential of TENS to convey different sensory cues according to terrain characteristics has not yet been explored. This study aimed to develop and test a novel encoding algorithm for lower limb sensory feedback restoration enabling the recognition of different terrain textures. The system performance was evaluated on five healthy participants for discrimination of three terrain textures. The proposed approach led to a correct classification of grass, stones and tiles in 88%, 81% and 79% of the cases with an overall accuracy of 83% outperforming literature results. Index Terms—Lower limb amputees, prosthetics, sensory feedback restoration, somatotopic sensations, Transcutaneous Electrical Nerve Stimulation I. INTRODUCTION THE lack of sensory feedback in current lower limb prostheses represents a major factor contributing to their abandonment [1]. In this field, Transcutaneous Electrical Nerve Stimulation (TENS) allows the elicitation of referred tactile sensations in a non invasive way since it uses superficial electrodes to electrically stimulate the underlying nerves. It has been applied in healthy participants and lower limb amputees improving gait symmetry, weight distribution and reducing phantom limb pain in the latter [2]–[5]. Nonetheless, although ground unevenness is one of the most influential environmental factors affecting the risk of falls in amputees [6], the possibility of eliciting different sensations via TENS according with floor conditions has not yet been investigated. Therefore, this study aimed to develop and test a novel encoding algorithm for lower limb sensory feedback restoration allowing different terrain textures recognition. The proposed system was tested on five healthy participants for three terrain textures discrimination. This study was supported by the Italian Ministry of Research under the complementary actions to the National Recovery and Resilience Plan (NRRP) “Fit4MedRob - Fit for Medical Robotics” Grant (PNC0000007). II. MATERIALS AND METHODS Five healthy participants (2M/3F, 23.2±1.9 years) were enrolled in the study. The experimental setup included: i) a sensorized insole endowed with six Force Sensing Resistors (FSRs) for estimating vertical Ground Reaction Force [5]; ii) the FSRs acquisition system and signal conditioning circuit housed in a 3D-printed case fastened to the thigh via velcro straps; iii) the electric stimulator STG4008 for delivering stimuli to the participants’ right leg’s tibial nerve through two circular (25 mm diameter) and superficial electrodes located near the popliteal fossa. The experimental protocol was divided into two phases. The first one aimed to the characterization of the sensations that can be elicited via TENS in the lower limb. For each stimulus, the participants were asked to indicate the location of the evoked sensations and describe them in terms of naturalness, location, intensity, pain and quality. The second phase aimed at evaluating participants’ ability to recognize three different terrain textures (i.e., grass, stones and tiles [7]) relying exclusively on TENS. The experimenter worn the instrumented insole and walked on the terrains: this choice ensured that terrain recognition depended solely on TENS feedback, avoiding any bias from tactile information from the participant’s plantar surface. The data acquired from the instrumented insole were used to modulate the electric stimulus delivered to the participant. The symmetric biphasic square wave in Fig.1 was adopted [2]–[5]. The Pulse Amplitude (PA) and Pulse Frequency (PF) were kept constant to PAMAX (i.e., a current value specific for each participant) and 500 Hz [2], [3], [5], respectively, the PW was varied along walking as follows PW(V)=P Wmin +P Wmax −P Wmin 21 + sin 2π(V−V0) Vmax −V0 (1) where PWMAX and PWMIN were the maximum and minimum values specific for each participant, respectively. V, V0 and VMAX indicated the current, minimum and maximum value of the sum of the readout of the six FSRs of the insole. During familiarization, the experimenter performed three walking trials for each terrain texture and the participant familiarized 2025 I-RIM Conference October 17-19, Rome, Italy ISBN: 9788894580570 10.5281/zenodo.17629862 223 with the elicited sensation. During validation, the experimenter walked ten times for each terrain in a random order and the participant, blindfolded and acoustically shielded, had to recognize the texture. The performance of the system was quantified in terms of: i) perceptual thresholds; ii) sensations descriptors; iii) terrain textures recognition accuracy. Fig. 1. Block diagram of the experimental setup for terrain recognition. III. RESULTS AND DISCUSSION The participants sensory and motor thresholds were 8.1±0.9 mA and 12.2±1.2 mA, respectively. The elicited sensations were mainly described as natural for charge modulation (51%) and almost natural (41%) for the frequency one. The stimulation evoked mostly superficial and painless sensations described as tingling or vibration: 27%, 22% and 26%, 43% for charge and frequency modulation, respectively. The results shown in Fig.2 demonstrated a strong linear correlation between both injected charge (Pearson correlation coefficient ρ=0.83) and stimulation frequency (ρ=0.82) with the perceived intensity. During charge modulation, the perceived area increased reaching over 60% of the sole, 50% of the back and 40% of the calf. A similar trend was encountered during frequency modulation, too. Such results are in line with those ones already reported in literature demonstrating the good performance of TENS in eliciting tactile sensations in the lower limb [2]–[5]. As far as the terrain recognition, grass, stones and tiles were rightly discriminated in the 88%, 81% and 79% of the cases, respectively, with an overall accuracy of 83% outperforming the literature (63% [7]). The tiles were misclassified mainly as grass (11%) since such terrain were characterized by comparable level of uneveness. Indeed, the even terrains (i.e., grass and tiles) were correctly discriminated in 93% of times, while the uneven one (i.e., stones) was rightly discriminated in 81% of the cases. IV. CONCLUSION A novel encoding algorithm for three terrain textures recognition via TENS applied to the lower limb was designed, developed and evaluated on five participants. The proposed stimulation strategy allowed the correct terrain textures classification with an accuracy of 83%. Future studies will be Fig. 2. A-B Relationships between the stimulus intensity and that one referred by the participants; CLocation of the elicited regions at the minimum, medium and maximum stimulation intensity during charge (red) and frequency (blue) modulation. Fig. 3. Normalized confusion matrices for grass (G), stones (S) and tiles (T) (A) and even/uneven (E/U) terrain (B) recognition. dedicated to the experimental validation with amputee subjects where the presence of fibrotic tissue and neuroplasticity could influence the discrimination capability. REFERENCES [1] Raspopovic, S. et al. (2021). Sensory feedback for limb prostheses in amputees. Nature Materials [2] Demofonti, A. et al. (2020). Somatotopical feedback restoration in the lower limb through TENS: a feasibility study. In Convegno Nazionale di Bioingegneria [3] Demofonti, A. et al. (2021). Modulation of sensation intensity in the lower limb via Transcutaneous Electrical Nerve Stimulation. 43rd Annual International Conference of the IEEE EMBC [4] Pan, L. et al. (2020). Evoking haptic sensations in the foot through highdensity transcutaneous electrical nerve stimulations. Journal of neural engineering [5] Demofonti, A. et al. (2025). Restoring Somatotopic Sensory Feedback in Lower Limb Amputees through Noninvasive Nerve Stimulation. Cyborg and Bionic Systems [6] Kendell, C. et al. (2010). Indicators of dynamic stability in transtibial prosthesis users. Gait & Posture [7] Prasanna, S. et al. (2023). Uneven terrain recognition using neuromorphic haptic feedback. Sensors 224