PROJECTS
(A) ADVANCED INSTRUMENTATION/SIGNAL PROCESSING FOR THE DETECTION AND INTERVENTION IN OBSTRUCTIVE SLEEP APNEA
Obstructive Sleep Apnea (OSA) is a serious sleep disorder caused by a complete or near complete collapse of the air passage, leading to a large number of untimely deaths. About 10% of all the snorers may have moderate to severe OSA, unknown to them, nevertheless requiring medical attention.
The standard method of diagnosis, known as polysomnography, requires that the patients spend one or several nights in the sleep-clinic of a hospital, wired to a multitude of instruments such as ECG/EEG recorders, chest bands, microphones and video equipment. Polysomnography is inconvenient and very expensive, making it unsuitable for long-term monitoring or mass screening purposes. In this project we attempt to develop portable instruments and signal processing techniques to detect/predict OSA in a home setting. We are currently focusing on identifying the minimal number of data modalities required to reliably detect OSA. The candidates under investigation include EEG, ECG, and, especially the snoring, which is a common symptom in OSA. The use of snoring signals is based on the premise that snoring carries vital information on the acoustical and structural properties of the upper airway, which collapse partially or fully during an OSA attack. The work in this project is being done in collaboration with the Sleep Clinic, Singapore General Hospital Pvt. Ltd., Singapore, Princess Alexandra Hospital, Brisbane and The University of Tokushima, Japan.
(B) FUNCTIONAL IMAGING OF THE NERVOUS SYSTEM AND THE DESIGN OF NEURO PROSTHETICS DEVICES
Injuries of the nervous system, especially of the brain and the spinal cord, bring in devastating consequences with only slight chances of complete recovery. Depending on the position and the nature of the injury, the patient may permanently loose basic functions such as breathing, the control of limbs, bowels, bladder and the like. Functional Electrical Stimulation (FES), a method in which a carefully controlled pulses of electrical energy is applied to appropriate muscles/nerves, is being considered as one solution to this problem. In this project we propose to: (a) capture propagating signals directly from nerve bundles using cuff electrodes and analyze them for intention of movement; (b) develop novel techniques to inject appropriate stimulation currents directly to nerves achieving selective stimulation of a given group of muscles; (c) to adapt techniques from functional brain mapping with to study the peripheral nervous tissue, and (d) to develop implantable neuro-prosthetic devices to restore/augment peripheral motor functions.
(C)ULTRASONIC IMAGING, IMAGE PROCESSING AND TARGET CHARACTERIZTION
Due to imperfections in imaging equipment and to the fundamental complications in the propagation of ultrasonic waves in tissue, the image displayed on a medical ultrasound scanner represents a distorted version of the true tissue information. These distortions can hide early stage (small) or low-contrast tumors, which, otherwise might be seen on a perfect imaging system. The early diagnosis of malignancies is of paramount importance in treating cancer, but it still remains a daunting problem. In this project we attempt (a) to identify the distortions associated with the imaging process and to nullify them, (b) to provide the radiologist with quantitative/objective measures describing the clinical states of tissue, and, (c) to develop parametric imaging and 3D visualization techniques.
(D) HIGHER-ORDER SPECTRAL TECHNIQUES IN BIOMEDICAL SIGNAL RECONSTRUCTION AND INTERPRETATION
Many biomedical signals have a complex structure requiring more sophisticated tools of analysis than the traditional autocorrelation and Gaussian approaches. Higher-order statistics (HOS) has received wide attention as a tool for the reconstruction of non-minimum phase signals because of its ability to preserve true phase and to reject Gaussian noise. A major disadvantage towards the exploitation of HOS techniques in medical signal processing is that they have a higher computational complexity compared to those of autocorrelation approaches. In this project, we attempt to develop low-complexity techniques for EEG signal reconstruction and interpretation based on HOS. The developed algorithms will be considered for DSP implementation.
(E) MUSIC, MEDITATION AND RELAXATION IN HUMANS.
We have obtained some basic results in the area of relaxation and the effects of meditation and music. Projects are available in the areas of instrumentation and Signal Processing, and application design.
Emotional/mental states of humans are often embedded in physiological signals/images such as the EEG (brain wave), ECG, EMG/the skin galvanic response (used in lie detectors), voice, and, facial expressions. Among the complex set of emotions displayed by humans, relaxation is relatively easy to detect and has many potential uses. It is well known that music and transcendental meditation can induce a 'relaxed-state' in humans, under appropriate circumstances. In this project, we plan to develop a relaxation-index (as computed through biosignals) and explore the possibility of using it in real-world applications. The project will involve experiment design, biosignal acquisition in the laboratory and signal analysis. Students interested in this project are strongly encouraged to meet Dr. Abeyratne for further details.
(F) INSTRUMENTATION AND SIGNAL PROCESSING IN UROLOGICAL APPLICATION
Projects are available in the areas of ultrasound imaging and electrophysiological signal acquisition and processing in urology (Paediatric and Gerontology applications)
