Personal tools

You are here: Home / Training / CDT in Healthcare Innovation / Student Research Areas / 2012 Abstracts

2012 Abstracts

Patient-Specific Modelling of Abdominal Aortic Aneurysm Evolution

Jesse Boamh, Linacre College

Abdominal aortic aneurysm (AAA) is a vascular disease, characterised by a localised and progressive enlargement of the abdominal aorta. AAA is known to rupture occasionally, and in that case the outcome is usually catastrophic. Overall, this disease accounts for 1-2% of all deaths in the UK. Up to date, there is no satisfactory methodology to assess whether a detected aneurysm is in risk of rupture and simplistic size correlations are used. Currently, I have attained a set of successive patient follow-up clinical scans for a detected aneurysm. Using image reconstruction techniques I have segmented the real geometry model of the detected aneurysm. I apply a fluid-solid-growth (FSG) computational framework to simulate the mechanobiology of the arterial wall of the patient-specific geometry of AAA. The model accounts for the structural arrangement and natural reference configurations of elastin and collagen of arterial wall components. The wall of the artery is subjected to a constant systolic pressure and physiological axial stretch.

The results emphasise the role of wall shear stress (WSS), elastin degradation, and collagen remodelling in AAA enlargement. Low WSS promotes elastin degradation as AAA enlarges. The largest aneurysmal growth is found at the region of low WSS (< 0.2 Pa). An asymmetric bulge is seen in the abdominal aorta region and aneurysm evolution enlarges faster at the upstream region. This yields greater insight into AAA evolution.

Strategies for Enhancing the Efficacy of Sonothrombolysis using Magnetic Microbubbles

Calum Crake, Lady Margaret Hall

Stroke is the third most common cause of death in developed countries and kills or disables 10 million people per annum. The majority of stokes are ischaemic in nature, caused by a blood clot in the brain. Thrombolytic drugs such as t-PA are the only approved treatment for ischaemic stroke, but must be administered within a few hours of onset and risk serious side effects; as such, 95% of patients do not receive any treatment.

Ultrasound and microbubbles have both been shown to enhance the effect of thrombolytic drugs. However, early results of in vivo and clinical trials have been disappointing, and found to pose considerable risk for the patient. Improved understanding of the mechanisms of ultrasound and microbubble-enhanced thrombolysis is required to devise safe and effective treatment regimes. In addition, improved targeting of treatment is necessary to reduce side effects, increase eligibility and improve outcomes. In particular, biophysical targeting methods such as the use of magnetically-targeted microbubbles may be used to increase the selectivity of treatment and further improve ultrasound-enhanced thrombolysis.

Non-Contact Remote Monitoring

Alessandro Guazzi, Kellogg College

Continuous vital sign monitoring, already possible through a number of contact technologies including photoplethysmography (PPG) and particularly pulse oximetry, allows a continuous assessment of the cardiac rhythm and the arterial oxygen saturation, which can be used as indicators of the overall health of the patient. Such a continuous surveillance is useful in the treatment of chronic diseases as it allows the early detection, and thus potentially improves the prevention, of catastrophic events. Pulse oximetry, recognised as one of the most important advances in patient monitoring in the past decades, has in fact become internationally mandatory for patients undergoing anaesthesia, and is thought to be a potential predictor for a number of complications in post-operative patients, including head trauma and airway obstruction. The detection of further parameters such as blood pressure and arterial behaviour has also been shown to be possible through the analysis of the PPG signal, and could lead to improved diagnoses and disease prevention in patients. The development of remote vital sign monitoring would retain all of the advantages of contact monitoring while also reducing the discomfort to patients and risks and costs related to the spread of infections in a hospital setting.

Breathing-rate estimation using video photoplethysmography

Joao Jorge, St Hilda's College

Breathing rate is one of the vital signs most indicative of critical illness and cardiorespiratory deterioration. The ability to assess respiratory rate remotely and track its changes over time would provide a valuable addition to existing health surveillance systems. Photoplethysmography (PPG) is a non-invasive optical technique that measures variations in skin blood volume and perfusion. The PPG signal contains components that are synchronous with both cardiac and respiratory rhythms. Typically, the optoelectronic setup of PPG has included a dedicated light source (e.g. at red and/or infra-red wave-lengths) but recent work has shown that pulse measurements can be acquired remotely using a consumer-level digital camera with normal ambient light as the illumination source. In light of these recent developments, the purpose of my research is to develop an accurate camera-based algorithm for measuring the respiratory rate of renal failure patients undergoing outpatient haemodialysis. If PPG proves to be a useful means for monitoring respiratory rate in this patient group it may be possible to design a single non-contact sensor to monitor breathing rate in addition to multiple other vital signs simultaneously.

Modelling the progress of neurodegenerative diseases

Simao Laranjeira, St Cross College

Neuronal degenerative diseases have been of great concern since their discovery. The number of patients that suffer from these pathologies is dramatic and, due to an aging population, these will tend to increase.  With modern medicine it is possible to care for patients for many years. Unfortunately these treatments incur large costs on health insurance systems. For all these reasons there is a lot of research surrounding them. In the research I’m involved with we are concerned with the role of the neurotransmitter glutamate. This is because the mechanisms in which this amino acid is involved are central for neural tissue function (i.e. memory and learning). However, it has been found that it is also toxic to the neuron when found in high concentrations in the extracellular space. This has already been identified as the main cause for neuron death during a stroke.

Our approach to studying the dynamics of this neurotransmitter is through the creation of a mathematical model that simulates cerebral tissue behaviour. It will have the potential to 1) forecast the progression of these diseases; 2) identify new biomarkers and 3) aid in the development of patient specific drugs.

Monitoring of Gestational Diabetes and physiological pattern prediction in pregnant women

Lise Loerup, New College

The first part of this project aims to test a telehealth solution designed to provide monitoring and support for women with GDM. Conventional monitoring of the condition is demanding and costly, as it involves intensive self-monitoring of blood glucose and frequent clinic visits. Evaluation of the system will be performed through a cohort study and randomized controlled trial, with aim of reducing time and saving costs for the patients and the health service, while maintaining a good level of care.

The second part of the project deals with prediction of physiological patterns in pregnancy. By recording vital signs (pulse, blood pressure, ppg, respiration, temperature) in a large population of pregnant women, healthy and pathological patterns and trends will be established.

Endothelial heterogeneity in the evolution of vascular disease

Aikaterini Mandaltsi, Wadham College

Aneurysms (excessive, permanent and localised dilatations of the arterial wall) present low rupture occurrence, but very high mortality and morbidity rates, when they rupture. While there are surgical interventions available, they are risky, costly and their evaluation is currently based on insufficient criteria. There is therefore a clinical need to develop a risk-of-rupture assessment, which can model the predicted evolution of an aneurysm.
The endothelium, the first arterial wall layer exposed to hemodynamic forces, translates mechanical stimuli from blood flow into physiologically important messages within the wall, resulting in remodelling of the wall structure in order to return the mechanical environment, for example the wall shear stress (WSS), to homeostatic levels. Endothelial cells show spatially and temporally varying properties, responding to the unique needs of the underlying tissue and to pathophysiological stimuli.

Endothelial heterogeneity, for instance, in defining WSS homeostasis, hadn’t previously been introduced in models of vascular disease evolution. Concepts, such as a temporally adaptive and spatially heterogeneous WSS homeostasis, implemented in 3D models for aneurysm evolution, can be further employed to model arterial adaptation in response to clinical intervention with devices as well as improve the development of tissue engineered vascular constructs.

Engineering ultra sound image analysis solutions for resource-constrained environments

Ali Maraci, St Hilda's College

Ultrasound (US) has been shown to be a safe and effective imaging modality in detecting pregnancy complications such as breech presentation. The non-invasiveness of this technique, alongside its cost efficacy and availability have promoted its uptake in the developed world for routine pregnancy scans and examinations. However the use of US is far less common in low income countries, particularly in rural areas, as there is a lack of training for effective use of this technology and accurate interpretation of the images as well as a relatively high cost being associated with the current US devices.

Recent technological advancements in the field of US have led to lower-cost and portable US devices, facilitating the use of US in the developing world. In light of the factors that can affect the quality of image interpretation, we have implemented a machine learning technique to segment the fetal head from images obtained from a low-cost USB US device so as to further facilitate the effective implementation of this device in developing countries. Fetal images from a high-end US machine were also obtained as a comparison to the USB US device. The results presented here illustrate that the algorithm works successfully on images obtained from both devices and that statistically there is no significant difference between the performance of the algorithm on the two.

Cardiac segmentation from Cardiac MRI

Tasos Papastylianou, Kellogg College

Cardiac segmentation from clinical images is an on-going area of research. Accurate and practical segmentation enables better quantification of existing clinical parameters, and has the potential to lead to better understanding of physiology, as well as pathology via new biomarkers of disease. The vision for the project is to define computational models of electromechanical activity of the heart, using MRI image analysis and personalised modelling techniques, to investigate recovery from myocardial infraction, particularly in the acute phase.

The current stage of the research focuses on the definition and refinement of an anatomical model, based on a probabilistic approach to cardiac segmentation; strengths and weaknesses of such an approach are investigated with respect to the existing literature. An approach to pure statistical classification using 2D or higher feature space for classification was attempted. This approach, involving pixel intensity and relevant distance metrics as features, with the addition of Markov Random Fields as spatial priors, showed a high number of false positive results for cardiac tissue. Further approaches will employ more feature-guided and anatomically-driven approaches, while retaining the probabilistic nature of the segmentation (such as probabilistic level sets), in order to exploit partial volume effects to build more accurate models from MRI.

Engineering Microbubbles for Quantitative Imaging

Paul Rademeyer, Pembroke College

Ultrasound imaging is a powerful real time modality that currently accounts for over a quarter of all clinical imaging procedures in the UK. The relatively low cost of the equipment, portability and patient safety is unique to medical imaging modalities. Progressive research into Microbubble Contrast Agents, encapsulated gas bubbles between 1 to 10µm in diameter, has shown the potential to enable quantitative ultrasound measurements. Contrast Enhanced Quantitative Ultrasound (CEQUS) imaging has the potential for accurate detection of tumours and cardiovascular diseases. However a significant challenge is the prediction of interaction between microbubbles, their environment and an ultrasound field. The characterisation of a microbubble’s shell properties is therefore essential for developing a model that can predict their behaviour accurately. My current work presents a design and feasibility study into a novel characterization method, combining simultaneous acoustic and optical measurements from unconstrained single microbubbles.

Perinatal Multisensor Monitoring System for Low-resource Settings

Lisa Stroux, Wolfson College

Foetal and maternal health is a major global health concern. Foetal mortality was estimated 2.5 million worldwide in 2008-2009, while 360.000 women were dying of pregnancy-related complications. Close to 99% of foetal and maternal deaths occur in developing countries. One approach to contribute to the prevention of avoidable deaths and disabilities is accessible perinatal risk screening and appropriate medical referral.

My research investigates the concept of an affordable, non-invasive and easy-to-use perinatal monitoring system for resource-constrained environments. A system is being built combining multiple low-cost sensors connected to a smart phone to enable signal processing. The concept focusses on the assessment of three indicators for foetal wellbeing; the foetal heart rhythm, foetal growth and foetal presentation. The foetal cardiac activity is sensed using a 1D ultrasonic probe; a maternal pulse oximeter is connected with the aim to cancel contamination introduced by the mother’s heartbeat using source separation techniques. The use of the in-built phone accelerometer is being investigated to locate the ultrasound probe, enabling the estimation of the foetal position based on the cardiac activity, and to derive the size of the foetus.  Primary development work will include the establishment of applicable signal quality metrics and appropriate signal processing algorithms.

Crowd-sourcing cardiac data using Bayesian voting in resource-constrained environments

Tingting Zhu, St Cross College

Cardiovascular disease currently accounts for 30 percent of global deaths and is predicted to remain as the single leading global burden of injury and death for the next 20 years, particularly affecting lower- and middle-income countries. Electrocardiogram (ECG) is the gold standard for cardiac abnormality screening. The lack of trained experts to provide accurate reading of the ECGs compounds the problem. Multiple noncolluding experts/annotators and/or algorithms can be combined to improve the accuracy of class labelling tasks. My work is focused on building a stand-alone ECG annotation system which consists of two parts: 1) Crowd-sourcing ECG annotations and provides training to non-experts to identify cardiac arrhythmia. 2) Act as a back-end server to an electronic medical healthcare record system and provides instantaneous diagnosis of cardiac arrhythmia by combining multiple annotators and/or algorithms using a Variational Independent Bayesian classifier combination framework.