Artificial Muscle Intelligence System With Deep Learning for Post-Stroke Assistance and Rehabilitation
| Autor: | Sunil Jacob, P G Vinoj, Leonardo Mostarda, Fadi Al-Turjman, Varun G. Menon |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
medicine.medical_specialty
General Computer Science Computer science medicine.medical_treatment assistivetechnologies 02 engineering and technology Electroencephalography 01 natural sciences Transcutaneous electrical nerve stimulation law.invention Physical medicine and rehabilitation law 0202 electrical engineering electronic engineering information engineering medicine Paralysis General Materials Science EEG BCI Brain–computer interface Rehabilitation medicine.diagnostic_test business.industry Deep learning exoskeleton 010401 analytical chemistry General Engineering healthcare Artificial muscle intelligence 0104 chemical sciences Exoskeleton Gesture recognition Post stroke 020201 artificial intelligence & image processing Artificial muscle lcsh:Electrical engineering. Electronics. Nuclear engineering Artificial intelligence medicine.symptom business lcsh:TK1-9971 Gesture |
| Zdroj: | IEEE Access, Vol 7, Pp 133463-133473 (2019) |
| ISSN: | 2169-3536 |
| DOI: | 10.1109/access.2019.2941491 |
| Popis: | Stroke is one of the prime reasons for paralysis throughout the world caused due to impaired nervous system and resulting in disability to move the affected body parts. Rehabilitation is the natural remedy for recovering from paralysis and enhancing the quality of life. Brain Computer Interface (BCI) controlled assistive technology is the new paradigm, providing assistance and rehabilitation for the paralysed. But, most of these devices are error prone and also hard to get continuous control because of the dynamic nature of the brain signals. Moreover, existing devices like exoskeletons brings additional burden on the patient and the caregivers and also results in mental fatigue and frustration. To solve these issues Artificial Muscle Intelligence with Deep Learning (AMIDL) system is proposed in this paper. AMIDL integrates user intentions with artificial muscle movements in an efficient way to improve the performance. Human thoughts captured using Electroencephalogram (EEG) sensors are transformed into body movements, by utilising microcontroller and Transcutaneous Electrical Nerve Stimulation (TENS) device. EEG signals are subjected to pre-processing, feature extraction and classification, before being passed on to the affected body part. The received EEG signal is correlated with the recorded artificial muscle movements. If the captured EEG signal falls below the desired level, the affected body part will be stimulated by the recorded artificial muscle movements. The system also provides a feature for communicating human intentions as alert message to caregivers, in case of emergency situations. This is achieved by offline training of specific gesture and online gesture recognition algorithm. The recognised gesture is transformed into speech, thus enabling the paralysed to express their feelings to the relatives or friends. Experiments were carried out with the aid of healthy and paralysed subjects. The AMIDL system helped to reduce mental fatigue, miss-operation, frustration and provided continuous control. The thrust of lifting the exoskeleton is also reduced by using light weight wireless electrodes. The proposed system will be a great communication aid for paralysed to express their thoughts and feelings with dear and near ones, thereby enhancing the quality of life. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|