Low-power machine learners for implantable decoding
Approximately 6 million people in the US and roughly 1 in 50 people worldwide suffer from paralysis. Intracortical brain machine interfaces (iBMIs) have shown promise in aiding movement, self-feeding and communication abilities of these severely motor-impaired patients. iBMIs essentially take neural...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis-Doctor of Philosophy |
| Language: | English |
| Published: |
Nanyang Technological University
2021
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/146463 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Approximately 6 million people in the US and roughly 1 in 50 people worldwide suffer from paralysis. Intracortical brain machine interfaces (iBMIs) have shown promise in aiding movement, self-feeding and communication abilities of these severely motor-impaired patients. iBMIs essentially take neural activity as an input, which is then subjected to signal processing and neural decoding, in order to drive prosthetics. However, the current systems are bulky, wired, immobile, conspicuous and require frequent calibration procedures often in the presence of a neural engineer. In this thesis, we have explored algorithmic and circuit and system level solutions to the aforementioned problems. Accordingly, we have presented circuit and system-level studies on offline and real-time non-human primate (NHP) data in order to aid development of scalable fully implantable wireless iBMIs. Furthermore, we have looked at novel algorithmic solutions to reduce calibration procedures. |
|---|