Neural engineering

Mechanical Engineering — the branch of science and technology concerned with machines, and things that move — and Neural Engineering dovetail nicely. The brain, after all, is simply a control system driving our limbs, speech organs, eyes, diaphragm, and more.

Decoding the EEG

In humans, sensory perception and voluntary movement are controlled by complex spatial and temporal patterns of brain activity. This activity can be measured by the electroencephalogram (EEG) which uses electrodes that are non-invasively and painlessly adhered to the scalp. EEG is potentially useful in brain-machine interfaces (BMIs), potentially enabling those with mobility impairments resulting from trauma to the peripheral nervous system to control external assistive devices, such as a prosthetic limb. To be useful, a BMI would need to be able to decode brain activity relating to both perception and movement (these decoded signals would then be used as control signals for the external device).

This project uses EEG to measure complex patterns of brain activity in normal subjects (specifically, responses of visual areas of the brain to visual stimuli). The overarching goal of the project is to develop new ways to decode these measured signals. Having measured a pattern of brain activity, can we use a decoder to determine what visual stimulus was shown to the subject.

Understanding the functional organization of biological neural networks

The brain is a complicated biological neural network (BNN). A chunk of brain the size of a match head contains thousands of neurons, each one receiving input signals from hundreds of nearby “upstream” neurons, and in turn sending its output signal to hundredsof “downstream” neurons, also nearby. This project concerns the use of machine learning to help understand the functional organization of BNNs.

We’re using optogenetics data previously collected by our colleagues, Drs Peter Freestone (FMHS) and Mark Trew (ABI). These data are electrophysiological recordings (ie., they’re recorded with a microscopic electrode) made in an area of the brain (of an animal model) that’s important to the control of voluntary movement. The data describe responses of single neurons to optical activation of their “upstream” neighbours. We’re using this data to train a machine learning algorithm (MLA) to reveal the functional organization of a BNN. We’ll then use the MLA to analyze new data, thus automating the process of experiment and analysis. This project is an important piece of a puzzle: how does the brain control voluntary movement, and what goes wrong in “movement” diseases like Parkinson’s?

Eye-tracking for use in diagnosing visual deficits

Eye-tracking can be used to diagnose visual deficits. For example, an individual with poor acuity in one eye is unlikely to be able to use that eye to follow a small, moving target as it transits their visual field. Eye-tracking may be especially useful in diagnosing patients unable to communicate verbally, such as infants. This work concerns the design, build and implementation of a computerbased test of a patient’s ability to track a moving target. We are developing ways to quantitatively analyze tracking data. Ideally, this analysis will be sensitive to certain visual deficits, and will also teach us something about the biological basis of those deficits.

Electromechanical system for precise stimulation of the sensory epithelium

The somatosensory epithelium, such as the human fingertip, is exquisitely sensitive to vibration, enabling us to perform an impressive range of object manipulations, and to interact with our environment generally. This work concerns the design of a piezoelectric system for use in experimental settings. The system is programmable to deliver precise vibrations with amplitudes ranging from 1 micron to 1 mm. The system will be used to measure, among other things, human sensitivity to finger tip vibration, and in turn to develop biologically plausible mathematical models of touch that may inform the design of robotics.

Understanding cortical microstimulation

The activity of neural networks in motor and somatosensory areas of the brain’s outer layer, the cerebral cortex, is important to the control of reaching, grasping, and manipulation. We are studying somatosensory networks in animal models, implanting, there, arrays to measure responses to (1) sensory stimulation, and (2) electrical stimulation of nearby cortex. This work may have translational potential, informing pre-clinical trials of novel sensorimotor neuroprosthetics in tetraplegic humans.

ENGGEN 770 Medical Devices
MECHENG 313 Design of Real-Time Software
MECHENG 736 Biomechatronics

• Dr Luke Hallum
• Dr Minas Liarokapis
• Associate Professor Andrew McDaid
• Dr Peter Freestone (Faculty of Medicine and Health Sciences)
• Dr Mark Trew (Auckland Bioengineering Institute)
• Dr Andreas Kempa-Liehr (EngSci)
• Mr Emanuele Romano