Abstract
The treatments for Parkinson’s disease, epilepsy and other brain disorders have relied mainly on medication, while medication is found to have degenerative or even adverse effects. Following the rapid development of integrated circuits, various implantable microsystems have been developed for neural rehabilitation. For example, neural activities of the motor cortex can be recorded for controlling external prosthetic devices. Deep-brain stimulation is also found useful for treating neural disorders such as the Parkinson’s diseases or epilepsy. However, most brain-machine-interface (BMI) microsystems have multiple recording channels share the same analogue-to-digital converter (ADC), and the recordings are processed in a digital core. This would hinder the possibility of stimulating neurons right after a specific pattern of neural activities is recorded. This thesis proposes the recording and stimulation circuits that facilitate closed-loop control on the neural stimulation. Pixel-level ADCs that convert neural recordings into digital pulses are used to replace the traditional ADCs. Each recording channel has its own pixel ADC whose pulse output can control the stimulators directly. A stimulator circuit that can set its output voltage automatically by feedback control is also designed. The proposed circuit would enable BMI microsystem to stimulate neurons in real time and in accordance with neural recordings. This feature allows brain to build extra connections between neurons. The proposed circuits have been designed and fabricated with the TSMC 0.18μm process. The measurement results are presented and discussed in this thesis.