Dr. Simon Kos
Innovation is at the heart of continuous improvement, which is how health raises the quality of care. The Three Horizons Model demonstrates a programmatic way of managing sustainable innovation, focusing on today, tomorrow and the future. The role of technology will be examined in three parts:

1) Optimising care today, using technologies to help communicate, measure, collaborate and connect.

2) Introducing new models for tomorrow, with technology supporting telemedicine, personalised healthcare, population health management and cloud computing.

3) Research for the future.

This will be done using local and international case studies of leading healthcare organisations. This model and these examples provide a structured approach to healthcare organisations seeking to improve care through technology innovation.


Dr. Marco Guermandi and Roberto Cardu
Localization of EEG signal sources is a well established method for providing low-cost, high temporal resolution brain activity maps. Since localization accuracy is deeply influenced by local conductivities inside patient's head, Electrical Impedance Tomography can provide priors to source localization algorithms which improve the reconstruction quality. In order to better integrate the two techniques, we are developing active electrodes containing an application specific integrated circuit with the purpose of providing a high-density system in which each electrode is configurable to perform either simultaneous EEG and EIT voltage readout or EIT current generation. In this presentation, we will introduce circuit design details, electrical performance characterization and some functional tests of the developed electrodes.


A/ Prof. Rosalind Sadleir
Functional MRI is a very commonly used method for identifying sites of neural activity within the brain. While fMRI signals are highly correlated with locations and times of activity, the BOLD contrast that gives rise to fMRI signals is a consequence of blood flow near active neurons, and is also time-delayed with respect to this activity. Many attempts have been made to image neural activity more directly using an MRI-based technique. Functional Magnetic Resonance Electrical Impedance Tomography (fMREIT) may have great promise for this application. The low-frequency conductivity of active cells changes during action potentials, due to changes in membrane conductances. Because conductivity is a scalar, it is not subject to cancellation. Other methods such as neural current MRI (ncMRI) that seek to image effects of magnetic field variations caused by activity, are predicted to only be capable of detecting activity in conducting axonal bundles. In this talk I will describe the mechanism of functional MREIT, and establish the feasibility of fMREIT for study of activity in both isolated neural preparations and in vivo. I will also demonstrate a volumetric model of active neural tissue that can be used to estimate MREIT signal sizes during different pulse sequences and current injection patterns.


Prof. Eung-Je Woo
Magnetic Resonance Electrical Impedance Tomography (MREIT) aims to produce images of electrical conductivity distributions inside the human body. We inject current into the body through surface or internal electrodes to probe the body. Injected current produces a distribution of current density, which is affected by the unknown conductivity distribution. The current density induces an internal distribution of magnetic flux density, which can be measured by using an MRI scanner. The conductivity image reconstruction problem in MREIT takes advantage of measured magnetic flux density data subject to multiple current injections. We describe the basic principle, mathematical and numerical modeling, experimental techniques and numerous experimental results. Potential clinical applications in tumor imaging and neuroimaging will be discussed.


Stephan Lau
While the effect of skull discontinuities on electric surface potentials has been addressed frequently, only very few studies exist on the influence of such discontinuities on the magnetic field. We investigated the effect of holes in the skull on the simultaneously recorded electroencephalogram (EEG) and magnetoencephalogram (MEG). Measurements were carried out in rabbits with two holes in the skull and an artificial current dipole positioned at various distances underneath one hole. Our results demonstrate that both EEG and MEG were significantly influenced by the holes with maximum MAGrel values of above 300% for EEG and above 20% for MEG. The strongest influence was observed when the source is under the edge of the hole. We conclude that discontinuities in the skull should be accounted for in volume conductor models used in the reconstruction of amplitudes of brain sources from EEG and MEG, particularly in infants, children and patients with post-surgical skull conditions.


Dr. Thomas Brennan
The burden of chronic disease or long-term conditions is dramatically increasing globally. The impact of chronic disease in low-resource settings is particularly severe due to the shortage of trained medical personnel and limited access to healthcare facilities. The main driver of this increased chronic disease burden is the rapid urbanization in low-resource settings, which is typically combined with poor nutrition and lack of exercise. The rapid uptake of mobile phone technology world-wide provides an opportunity to increase access to healthcare services in low-resource settings by integrating low-cost sensors with mobile phones to enable remote screening through a combination of self-monitoring and the deployment of health workers with very basic training in these settings. This talk outlines four new technologies developed at Oxford’s institute of Biomedical Engineering that effectively turn a mobile phone into a stethoscope, a spirometer, a blood-pressure monitor and a sleep monitor. The cost of the self-monitoring device is greatly reduced by embedding the signal processing on the mobile handset or in the cloud and using the battery of the phone as the power supply thereby reducing the cost of the ‘medical device’ as well as enabling longitudinal condition monitoring.


Prof. Hong Yan
The amount of molecular biological data, such as DNA and protein sequences, is increasing at a very rapid rate and the analysis of these data to understand the underlying biological structures and processes is a great challenge. In this talk, I will provide an introduction to signal processing and pattern recognition methods for biological data analysis, including spectral estimation, image segmentation, the Hough transform and spatial tessellation algorithms. I will present our recent work on using these techniques for gene expression data classification, gene regulatory relation inference, disease diagnosis, drug therapeutic effect assessment, and prediction of biomolecular interactions.


Dr. Slade Matthews
1. Non-chronotropic component changes in parasympathetic heart rate variability: a sensitive marker for perceived examination stress in healthy university students.
Heart rate variability (HRV) analysis reveals characteristics of autonomic regulation reflecting the balance between sympathetic and parasympathetic nervous outflow. Emotional stress affects sympathetic / parasympathetic balance and is reflected in heart rate variability changes. We sought to compare HRV with another sensitive measure of sympathetic nervous system activity – salivary cortisol – and a questionnaire based measure of perceived stress in a student cohort during exam time, relative to term time. Perceived stress and LFnu increased while PNN50 and HFnu decreased during exam time relative to non-exam period, consistent with hypotheses. Interestingly, no significant differences were observed for salivary cortisol or heart rate.

2. Morphometic analysis of oligodendroglioma histopathology for identification of phenotype differences
Morphometric analysis of histopathological images is rapidly emerging as a valuable diagnostic tool for a variety of diseases. A correlation in the literature has recently been observed between chemosensitivity and co-deletion of the 1p/19q chromosomal arms in oligodenrogliomas. Furthermore examination of expert classification of oligodenrogiomas has revealed strong relationships between morphological characteristics and phenotype. We aim to develop fast, affordable and objective diagnosis of oligodendrogliomas through machine learning and image analysis techniques to assist in the management of this tumour type.


Prof Paul Keall
Major technological innovations in cancer radiotherapy- used to treat 40% of cancer patients- have had a major impact on treatment outcomes through improved survival and reduced treatment-related toxicity. Even with these technological advances, a major unsolved problem is to image and treat tumours that move during treatment, particularly due to respiration. This work will cover the advantages and disadvantages of solutions to improve the current state-of-the-art, through audio-visual biofeedback, respiratory modulated imaging, real-time tumour position measurement/inference and real-time radiation beam/tumour realignment.


Dr. Gari Clifford
Noise contaminates biomedical recordings frequently, disrupting patient care and leading to distrust and confusion. As many as 95% of critical events requiring human intervention are false. In part this is due to the lack of intelligent interaction between physiological parameter streams. In resource-constrained locations there is an additional issue of lack of trained healthcare workers. We present an automated method to analyse events which discovers relevant features in multiple physiologic streams and fuses them together with signal quality indices. The system can teach users why the alarm was suppressed or not, and can learn from experienced users to improve accuracy. An example application of neonatal apnea is described which demonstrates that 89% of false alarms in an unseen test set can be suppressed. We also present an example in fetal and adult ECG monitoring to provide low cost point of care devices in resource-constrained locations.


Prof. Brian Day
The vestibular organs in the inner ear act as inertial sensors that signal the head's linear acceleration and angular motion. This information is used by the brain for many functions including the control of balance, navigation, voluntary movement and gaze. Apart from gaze control, it is not feasible to study the vestibular contribution to these functions using physiological stimuli since forced head movements interfere with the motor function under investigation. Galvanic vestibular stimulation (GVS) uses percutaneous electric currents to activate the vestibular system non-invasively without moving the head and therefore provides a means of studying central vestibular processing during a variety of actions. As well as a physiological probe, GVS also has potential for assessing the integrity of vestibulospinal pathways in pathological conditions. For example, by stimulating the left and right labyrinths independently it is possible to show lateralised asymmetries arising from unilateral peripheral lesions (e.g. schwannoma) or central lesions (e.g. stroke). The newer and related technique of stochastic vestibular stimulation (SVS), in which the stimulating current has a noise rather than rectangular waveform, may prove to be particularly useful for studying patients because the stimulus is less destabilising and reliable responses can be obtained in a shorter time.


Prof. Craig Anderson
Traditionally, the perception of neurology is that of a descriptive discipline, offering careful diagnostic assessment but few therapeutic options. However, the last two decades has witnessed considerable advances in clinical neurology, transforming it to very much an active discipline. In particular, development of sophisticated imaging technologies has radically increased the diagnostic certainty of a wide range of conditions in neurology. Where once neurologists had to incrementally improve their diagnostic certainty with post-mortem examinations long after a clinical encounter, there are now a range of investigations that allow in vivo detection of both the location and pathophysiology of lesions, and thus a window of opportunity for therapeutic intervention to reverse an underlying pathological process. The growth in computerised tomography (CT), magnetic resonance (MR), ultrasound and other imaging techniques has dispelled much of the mystery of the neurological examination, now allowing better tailoring of treatments and appropriate management decisions for patients. Stroke, as the 4th leading cause of global disease burden, is one of the most common neurological emergencies. Most (~80%) strokes are ischaemic in nature caused by acute occlusion of an artery leading to immediate reduction in blood flow within the corresponding cerebrovascular territory. The size and site of the occlusion, and the efficiency of compensatory collateral blood flow, determine the extent of impaired blood flow and resulting neurological symptoms from 'at risk' ('ischaemic') and/or dead ('infarcted') brain. Early spontaneous re-canalisation may occur from the endogenous release of tPA, a serine protease of the fibrinolytic system which converts the zymogen plasminogen into the active protease plasmin, leading to cleavage of fibrin and the dissolving of newly formed thrombin 'clot'. However, for most patients, and particularly in those with large occlusions, this natural physiological function is inadequate to avoid the outcome of infarcted cerebral tissue from the occluded vessel. Modern therapy for acute ischaemic stroke is based on the premise that early vessel re-canalisation and reperfusion are essential for the preservation of brain function and promotion of satisfactory outcome.

This talk with outline current management using neuroimaging in the hyperacute setting of ischaemic and haemorrhagic stroke, and introduce two ongoing investigator initiated and conducted clinical trials investigating novel treatment strategies in these conditions.


Dr. Terry Chilchott et al


Click to view a larger image

Biosensors decipher biological information at the molecular level and translate it into presentations suitable for dissemination by diagnosticians. The sensing involves contact of an aquatic biological sample with a substrate sensitized to a specific type of biological molecule and the transduction of the targeted interaction to a signal suitable for deciphering. Transduction to an electrical signal facilitates utilization of semiconductor technology for miniaturizing the unary sensing unit; 'large scale integration' of many molecularly programmed units into a biosensor array; and connectivity to informational technology for computing comprehensive presentations from such arrays. Electrical transduction further facilitates the utilization of electrical impedance spectroscopy (EIS) for extracting the bioinformatics as well as electrochemical and structural characterization of the basic biosensor. High-resolution EIS studies of various organic monolayers and peptides on silicon will be presented that illustrate the scope of EIS in characterizing the structural integrity, biocompatibility, oxidative vulnerability, electric connectivity, bio-electronic amplification capability and bio-functionality potential of these constructs as biosensors.

Figure 1: Inphaze high-resolution electrical impedance spectrometer and characterization chamber illustrating (a) the three-terminal configuration for characterizing (b) functionalized nanostructures on substrates such as (c) gold, (d) silicon, and (e) and (f) silicon dioxide. (a) A gearing mechanism regulated by springs (not shown) lowers a gasket of rectilinear cross-section onto functionalized surfaces and delicately seals precisely defined active areas. Electrolyte is perfused into the inner space of the chamber, wetting the active surface as well as the counter and reference electrodes. The working and counter electrodes inject a sinusoidal current of amplitude, i, and angular frequency, ω, through the surface. The recessed reference electrode monitors the voltage response of amplitude, v, and phase, ø (referenced to that of i) yielding the impedance magnitude (v/i) and phase ø. (c, d, e, and f) A spectrum of measurements is modeled by a series of electrical circuit elements, each comprising conductive (g's) and capacitive (c's) components reflecting properties of sub-structural layers of the functionalized surfaces.


Dr. Graham Brooker
Biomechatronics honours theses at the University of Sydney
This presentation examines the wide range of honours theses with a biomechatronics flavour that I have supervised. These range from a low cost pneumatic muscle controlled orthotic to improve hand grasp strength and an improved pegboard test for at home PD evaluation, to a Fleisch pneumotachograph made from syringe needles, and a whole lot of stuff in between.