CDU’s Biomedical Research Laboratory aims to find innovative solutions to some of the critical research questions in the fields of biomedical engineering and health informatics. The laboratory is used by a number of enthusiastic research staff and students, working on a range of research problems. Below is a short description of our facilities and key research themes.
The laboratory provides access to highly sensitive and accurate devices and equipment for different tests and research activities, including an Electroencephalogram (EEG) data acquisition system, utilizing a G.Tec USB biosignal amplifier (Guger Technologies OG, Austria), G.Tec g.GAMMASYS active electrode system and G.Tec g.GAMMAcap. For auditory evoked potential, ER.2 insert earphones are used using an external sound card (Creative Sound Blaster Omni Surround 5.1) together with the EEG acquisition system. Several sensors are available for studying the responses from human subjects to different stimuli and triggers.
The “Cocktail Party” effect can be described as the ability to interpret sound coming from a single source while spectrally masked by background noise. Our brain utilise the binaural cues available in the signal to enhance the signal to noise ratio enabling us to understand what a person wishes to listen to. Binaural hearing is the mechanism developed in the auditory pathway during infancy that allows us to listen to speech masked in background noise. Binaural hearing is basically the ability to combine the information from both ears to detect the location of a sound source or to distinguish a sound from background noise. It may be impaired in people who have suffered from prolonged hearing loss as children, for example, due to otitis media. Studies have shown that indigenous children are five times more likely to be diagnosed with Otitis media (OM) than non- Indigenous children. As hearing is associated with the brain and the auditory cortex, it is important to understand the brain responses as well when studying hearing loss and related disabilities. Auditory evoked potentials (AEPs), the brain’s response to auditory stimuli, may be used as an objective measure of hearing and auditory processing.
This research aims to develop, test and evaluate an approach towards building an objective methodology to detect binaural processing in the human brain using auditory EEG evoked potentials. Current research areas include:
- Applicability of Fast Fourier Transform in analysing the Auditory Evoked Potentials
- Evaluation and analysis of Auditory Evoked Potentials by mathematically modelling the brain response to the auditory stimulus
- Application of wavelet analysis on the Auditory Evoked Potentials in EEG signals
- Evaluation of Auditory Evoked Potentials from EEG, using Principal Component Analysis
- Reliability of Independent Component Analysis technique on Auditory Evoked Potentials
- Neural networks applicable in analysing the Auditory Evoked Potentials in EEG signals with limited sets of experimental data
Development of signal processing algorithms for the Electro Cardiogram
Traditionally biomedical signals, such as ECG signals, are analysed in the time- domain by skilled physicians. However, abnormalities may be subtle and difficult to detect by visual inspection. Pathological conditions may not always be obvious in the original time-domain signal. The analysis of biomedical signals is complicated by the fact that the signals are typically both highly irregular and non-stationary. Time dependent variations and transient phenomena play an important role. Traditional signal processing techniques, such as Fourier analysis, do not deal very well with these transient phenomena. Methods, which allow a more accurate local description and separation of signal characteristics, such as wavelet transforms may be more suitable. The purpose of this research is the development of signal processing algorithms to extract useful information ECG signals. We intend to develop methods, which provide a quantitative evaluation of particular signal features and classify particular patterns. The overall goal is to develop signal-processing tools, which are useful for medical diagnosis.
Machine learning in health informatics
Machine Learning has opened up a plethora of opportunities in Health Informatics research, where versatile applications of general Machine Learning and Deep Learning have the potential of making a major difference. Current research areas include:
- Breast Cancer detection
Breast cancer is the most common type of cancer affecting women worldwide. Various researchers have proposed methods and tools based on Machine Learning and Convolutional Neural Networks for assessing mammographic images, but these methods have produced detection and interpretation errors resulting in false-positive and false-negative cases. This problem can potentially be resolved by implementing effective image pre-processing techniques to create training data for Deep-CNN.
- Prediction of Cardiovascular Disease
Cardiovascular diseases (CVD) are among the most common serious illnesses affecting human health. CVDs may be prevented or mitigated by early diagnosis, and this may reduce mortality rates. Identifying risk factors using machine learning models is a promising approach. This research aims to build a model incorporating deep learning architecture to achieve effective prediction of heart disease.
Research Coordinator: Prof Friso De Boer