[en] Brain functions can be severely altered by accidents leading to traumatic brain injuries or by diseases such as strokes, cancers or Alzheimer's. Currently, there are no medical instruments capable of restoring cognitive function, but recent progress in neurophysiology coupled with development of engineering techniques has introduced the era of a new class of devices: neural prostheses or brain-computer interfaces (BCIs). BCIs carry great promises: possibilities to stimulate areas of the brain to prevent or treat failures of the nervous system. Today, most devices used in this domain are not capable of restricting their action to specified anatomical targets due to the large size of the electrodes. In 2007, the BES-group (imec, Leuven) developed silicon-based multi-electrode neural probe arrays interacting at a single neuron level (single-unit) for in vivo selective neuronal recording and stimulation.
My Master Thesis was developed with imec, Leuven, and focused on the realization of a multi-channel recording system for neural applications to enable a closed-loop approach (recording - stimulation - recording) that would allow for adjustment and fine tuning of the required stimulation pattern. The system’s read-out conditions and digitizes eight channels and provides the digitized output to an existing TI MSP430 based microprocessor system for wired or wireless data handling. The resulting platform can be used as a prototype for extensive experimental testing by biomedical scientists from the BES-group (imec).
The first step in the study is the analysis, from a theoretical standpoint, of the conditioning and digitalization of the signals: low-noise amplification, filtering, offset compensation and digitization. A signal-to-noise study is performed, leading to the selection of electronic components suitable for my project. In the second part of the work, the PCB layout is conceived in a miniaturized SMD design style. The code implemented in the microprocessor MSP430 is detailed both for the recording of a single channel and eight channels concurrently. Adaptation of the amplification factor can be carried out either by digital processing or by user interaction. Design of the wireless link between a PC and the board is then considered. The link is built from the eZ430-RF2500 Development Tool (Texas Instruments), a complete USB-based wireless development tool providing all the hardware and software to create a wireless network.
Eventually, the performance and limitations of the design are evaluated. The system handles properly the entire frequency range of action potentials (from 100Hz to 6kHz) for both wired and wireless configurations and shows very good results for signals with amplitude higher than 1mV_pp. Gain variation is available with either a stand-alone digital processing method or with user's control through keyboard and achieves satisfactory results for both accuracy and rapidity. The major limitation is a lack of accuracy for signals below 500〖μV〗_pp. Solutions for future designs are proposed including suppression of the offset compensation circuit and acceleration of the communication with the computer. Ideas for the addition of stimulation functions are proposed in hopes of a closed-loop approach. Last but not least, power supply and dissipation considerations are tackled.
