UCL WIKI

UCL Logo
Page tree
Skip to end of metadata
Go to start of metadata

neuromodulation device to Multi-Modality imaging phantom projectimaging of skin reddening Respiratory motion models from unsorted 4DCTAnalysis and Characterisation of Advanced LEDs for High-Density Brain ImagingAnalysis and Characterisation of Advanced LEDs for High-Density Brain ImagingHenry Lancashiretranscutaneous electrode design andZuzanaobust and automaticranscutaneous electrode design and assessment for a neuromodulation device to control urinary incontinence

Supervisors: Anne Vanhoestenberghe and Sean Doherty

One of the lesser thought of consequences of damage to the body’s nervous system is the impact on bladder function, with overactivity of the bladder musculature, poor coordination and loss of voluntary control.  Combined these lead to urinary incontinence. By targeted transcutaneous stimulation of peripheral nerves, it is possible to control the bladder and reduce incontinence. One of the major obstacles faced In developing a medical device for chronic use is that current electrode designs are not suitable for sensitive areas of skin that may become wet.This project will seek to address this issue. The student will review current electrode design types and be challenged to come up with innovative design ideas. These designs will be reviewed and narrowed down, the student will aim to produce a set of prototypes. They will design an experiment involving healthy volunteers, in which they will test their prototypes against an array of outcome measures. This experiment will use the sensitive skin on the inner forearm as the target location.This project would suit a student who is motivated by real-world application, interested in materials design and interested in learning more about designing and running trials with human volunteers.  The work will take place in the Aspire CREATe lab, on the Royal National Orthopaedic Hospital campus.


Alzheimer’s disease Mineralomics

Supervisors: Sergio Bertazzo, Elena Tsolaki

Student: Benjamin Baxter            

Pathological mineralisation processes are associated with a range of diseases including atherosclerosis, rheumatic fever, breast cancer and ovarian cancer, Parkinson’s and Alzheimer’s disease.Mineral deposits have been observed in many structures present in the brain. Mineralization of the basal ganglia which has been considered to be a result of aging for many years, however noted that large quantities of mineral deposits are correlated to cognition impairment and psychosis. Mineral deposits in brain tissue even though a very common phenomenon is yet not very well understood with fundamental information missing on the processes involved.   

       In the proposed project, we will focus first on physical characterisation of the mineral deposits found in basal ganglia. Samples to be analysed have been obtained from Alzheimer’s disease patients, elderly (>60 years old) and young (< 40years old) donors. The first part of the research has revealed a range of structures in the Alzheimer’s and elderly brain. Building on these results, this study aims to quantify the size and spatial distribution of the structures observed and also obtain quantitate data on the chemical composition of the structures in order to identify any differences between the structures observed in the two groups. This information will add to our understanding on brain mineralisation and bring us one step closer to understanding its mechanisms of formation. This innovative research will build on the comprehensive skill set of the applicant and take it in a new and exciting direction that focuses on truly medically important breakthroughs, such as the role of brain mineralisation in Alzheimer’s disease. 

           The student will participate in the development of the project, sample collection and preparation, experimental measurements, analysis of the data and interpretation. Initially, the project will make extensive use of electron microscopes, with new techniques potentially added as the research work progresses. This is an experimental research project where laboratory experience is a plus but not a requirement.


Developing flexible and highly directional optical acoustic generators

Supervisors: Erwin Alles and Richard Colchester

Student: Thomas Coady

Optical acoustic generators, where ultrasound is generated using light through the photoacoustic effect, have recently been shown to be a viable alternative to conventional piezoelectric technology. Efficient optical acoustic generators typically consist of a highly optically absorbing dye, which converts incident light into heat, and an elastomeric component that efficiently converts this heat into a pressure increase; typically polydimethylsiloxane (PDMS) is used due to its high thermal expansion coefficient. In addition, the easy handleability of PDMS allows for a wide range of manufacturing techniques resulting in applications of optical ultrasound that are hard to achieve using piezoelectric technology. The aim of this project is to optimise previously developed methods to fabricate highly directional optical ultrasound generators on the tip of a flexible optical fibre. In addition, the aim is to combine these generators with a novel optical excitation scheme to further maximise their performance. As such, the project is mostly experimental in nature; some prior lab experience would hence be advantageous but not essential.


Ultrasound scanner – teaching tool

Supervisors: Rebecca Yerworth and Jem Hebden

Student: Xu Zhao

Medical physics and Biomedical Engineers need to understand how medical imaging devices work, and investigative learning is beneficial. However, in clinical devices the physics/engineering is hidden from view. What is needed is a striped down device where the individual stages can be controlled by the students. During this project, you will need to acquire a detailed understanding of the principles of clinical ultrasound machines so as to design, build and test a prototype classroom ultrasound demonstration kit, including test phantoms, using low cost components. This project builds previous final year projects: So far, an ultrasonic range finder, aimed at the robotics market, shown its potential for this job when combined with an Arduino – it is your role to turn the vision it into reality.This project will suit a student who enjoys practical work and writing Arduino code.


Numerical integration of differential phase contrast images

Supervisors: Marco Endrizzi and Fabio Vittoria

Laboratory-compatible X-ray phase contrast imaging techniques generate images that are proportional to the gradient of the phase shift imposed by the investigated sample to the wavefront. It is very often the case that the information needed is the actual phase shift, and therefore a numerical integration is necessary before these images can be interpreted or subsequently processed by other data analysis algorithms. Although conceptually simple, numerical integration is a challenging problem when images are affected by noise. This project focuses on the understanding, characterisation and further development of differential phase images integration methods. The problem will be tackled from a theoretical point view at the beginning, with the aid of existing simulation software and numerical tools. Standard methods will be implemented and evaluated, followed by testing of new algorithms and procedures such as the Kalman filter. Experimental data will be available for a final benchmarking of the project's results against real datasets


EIT data analysis

Supervisors: Rebecca Yerworth and Richard Bayford

Electrical Impedance Tomography (EIT) is a monitoring technique, which is increasingly being used to monitor respiration related air distribution in neonatal lungs (http://cradlproject.org/).  In this project you will analyse some of the clinical EIT data collected during CRADL observational study to assess the effects of different reconstruction or data analysis methods to establish whether there is a significant difference in the derived clinical parameters (e.g. centre of ventilation). Most of the analysis code will be provided to you, but you may need to adjust it or write additional components. Therefore, experience with Matlab is desirable, but not essential.


Evaluating a very-low-cost commercial optical device for LIDAR imaging.

Supervisors: Jem Hebden and Pablo Perez-Tirador

Student: Mihir Rao

UCL’s Biomedical Optics Research Laboratory has designed and built sophisticated and expensive medical imaging systems based on measuring the flight times of near-infrared photons as they travel through tissue. However, new technology developed for the mobile phone market provides an inexpensive means of measuring distances using the time it takes for photons to travel between an optical source and a detector. The objective of this project is to characterise the sensitivity and spatial accuracy of the low-cost device, and investigate whether it can be used to generate simple images which reveal the locations of objects in a manner analogous to RADAR (the optical equivalent is known as LIDAR) and diagnostic ultrasound. The student will perform a series of experiments to determine the sensing capabilities of the device, and then build a simple LIDAR system. This project is most suitable for a student who enjoys building mechanical and/or electrical devices and has very good manual skills.


Characterisation of material for lung phantom

Supervisors: Rebecca Yerworth, Eve Hatten & Richard Bayford with Serena De Gelidi

Student: Karen Hsu

Physical phantoms are often used to test prototype devices in a controlled environment, and for calibration of clinical systems. Electrical Impedance Tomography (EIT) is a monitoring technique, which is increasingly being used to monitor respiration related air distribution in neonatal lungs (http://cradlproject.org/).  This project involves designing and implementing experiments to test the mechanical and electrical properties of conductive materials, including their durability, and compare the suitability of the materials for the generation of a conductive phantom of the neonatal torso. This project will involve use of the Instron Mechanical testing instrument in the Biomedical Engineering lab.  


Comparing phase retrieval methods; how to maximise image quality in x-ray phase contrast imaging?

Supervisors: Charlotte Hagen and Gibril Kallon
X-ray phase contrast imaging (XPCi) stands for a class of radiographic imaging techniques, which, in addition to x-ray attenuation, are sensitive to phase shifts. This is especially beneficial for the imaging of weakly attenuating biological tissues. In an XPCi radiograph, phase and attenuation information are typically present simultaneously. In order to quantify the phase contrast, or as a pre-requisite to performing 3D imaging via computed tomography, it is required to apply a phase retrieval algorithm to the data. At UCL, we have developed several such methods, relying either on the acquisition and combination of several XPCi radiographs or a series of simplifying assumptions, respectively. An open question is: Which of these methods provides the best image quality, especially at low photon numbers (i.e. low radiation doses)? The student will aim to answer this question by comparing the results of all methods when applied to XPCi data. This will directly inform the use of XPCi in several medical and biomedical areas requiring that a high image quality is achieved at a low radiation dose, e.g. tissue engineering, and in-theatre imaging of excised breast tissue. The project will be based on simulations (a simulation code implemented in Matlab will be provided), with a possibility of verifying results experimentally. Skills required:
knowledge of medical imaging with ionising radiation, image quality metrics, image analysis, programming in Matlab


Investigate the derivation of a marker of mitochondrial pressure passivity that is related to outcome in infants with hypoxic-ischaemic encephalopathy

Supervisors: Ilias Tachtsidis and Subha Mitra 

Student: Wisdom Ohia

Perinatal cerebral hypoxic-ischemia (HI) is a condition resulting from reduced oxygen delivery and blood flow occurring either in around the time of birth. It occurs in 1 to 2 per 1000 live births and can result in significant mortality and morbidity. Although significant advances have been made in the understanding of changes in brain physiology during this period and in the treatment of asphyxiated babies, the pathophysiological changes are yet to be fully explored. It is extremely important to assess the changes on brain perfusion and metabolism following HI to develop future neuroprotection strategies. This project is part of an exciting collaboration between the neonatal neurology and neuroprotection group in the neonatal unit in UCL/UCLH and the Biomedical Optics Research Laboratory at UCL that aims to deliver novel measurements to investigate brain physiology and neuromonitoring through combination optical (broadband near infrared) technologies along with magnetic resonance imaging and spectroscopy biomarkers. Both these groups are world leaders in their respective fields and is continuously contributing towards the advancement of optical neuromonitoring methods and future neuroprotection.We have recently published [1] that using our novel broadband NIRS derived metabolic reactivity index, we can identify infants with severe injury at the cot side early in life. It would be interesting to review the same dataset to identify the minimum period of recording that will give us the same identification of severity of illness. We would also like to include our off line data processing technique (wavelet analysis) into the online live monitoring software.  Any previous experience with MATLAB software will be an advantage.Reference (further reading)1. Mitra et al. Pressure passivity of cerebral mitochondrial metabolism is associated with poor outcome following perinatal hypoxic ischemic brain injury. Journal of Cerebral Blood Flow & Metabolism (2017)


Calibration and characterisation of 3D photogrammetry

Supervisors: Adam Gibson and Charlie Willard

Agisoft Photoscan is commercially available software that allows the user to render the 3D surface of an object from multiple photographs. In this project, the student will familiarise themselves with the principle of photogrammetry and investigate its role in medical physics. Potential areas of investigation include: (1) Recommendations for reliable, efficient photogrammetry; (2) photogrammetry of skin surface; (3) spatial characterisation and calibration using a scale bar; (4) investigation of other potential applications of photogrammetry. The student will gain familiarity with scientific photography and the use of Photoscan, and also write their own analysis software in Matlab. Previous experience of photography and Matlab would be valuable but are not necessary. 


Multispectral imaging of skin reddening

Supervisors: Adam Gibson and Cerys Jones

Student: Azraf Chowdury

Sunburn and radiotherapy both cause reddening of the skin. We have a multispectral imaging system which acquires photographs at a range of wavelengths. In this project, the student will use this system to characterise skin colour before and after insults such as warm water, cold water and Deep Heat cream (used for muscle pains). This project requires experimental skills and image processing, and will involve some computer programming. Previous experience of photography and Matlab would be valuable but are not necessary. 


Development of a novel approach to broadband NIRS spectra analysis in the image space using machine learning.

Supervisors: Ilias Tachtsidis, Zuzana Kovacsova and Joshua Russell – Buckland.

Student: Bogdan Elefteriu

Near-infrared spectroscopy (NIRS) is an optical technology that uses light to measure oxygen content in body tissue. It shines near-infrared light (600- 1000 nm) into tissue and measures how much light is transmitted through, which depends on the amount of oxygenated and deoxygenated haemoglobin in the tissue. Tracking the absorption of light enables to assess the oxygenation state of the monitored tissue. NIRS is non-invasive and easy to use, it is increasingly used in hospitals to help monitor patients in intensive and neonatal care. Broadband NIRS is an advanced technique that used hundreds of wavelengths simultaneously and enables the measurement of absorption spectra of tissue. Whereas current methods extract oxygenation information from the absorption spectra directly, we propose the use of image processing methods to investigate image features that relate to different oxygenation levels.
The task of the student would be to analyse broadband NIRS absorption spectra acquired in phantom measurements; to transform them into 2D images (e.g. through the calculation of distance matrices), derive image features and explore classification methods, e.g. machine learning. Due to the computational nature of the project, previous Matlab or Python experience is desired.


The implementation of real-time tissue oxygen saturation calculation in a broadband NIRS system

Supervisors: Gemma Bale and Zuzana Kovacsova

Student: Annie Gheasuddin

Near-infrared spectroscopy (NIRS) is an optical technology that uses light to measure oxygen content in body tissue. It shines near-infrared light (600- 1000 nm) into tissue and measures how much light is transmitted through, which mostly depends on the amount of oxygenated and deoxygenated haemoglobin in the tissue. Tracking the absorption of light enables to assess the oxygenation state of the monitored tissue. NIRS is non-invasive and easy to use, it is increasingly used in hospitals to help monitor patients in intensive and neonatal care. miniCYRIL is a miniature NIRS system for hospital use that uses hundreds of wavelengths (a technique called broadband NIRS) to measure oxygenation and metabolism. It was designed for cotside use in the neonatal intensive care to monitor brain injured newborns and help with diagnosis. While the current system only measures oxygenation and metabolism changes, novel NIRS absorption spectra analysis methods based on the use of light diffusion models enable the calculation of absolute oxygenation levels. These can be used for an immediate assessment of brain health.
The student’s task will be to implement a real-time calculation of tissue oxygenation within miniCYRIL. This will also require the development of additional hardware components, such as a reference spectrum acquisition tube. The new algorithm will be eventually tested in adult volunteer studies.
Matlab experience is desired.


Analysis and Characterisation of Advanced LEDs for High-Density Brain Imaging

Supervisors: Rob Cooper and Hubin Zhao

Student: Neera Senthivel

Our research group is developing wearable brain imaging devices that use light to non-invasively monitor human brain activity and be used to produce high-quality 3D images of brain function. This technology paves the way for investigating the brain of new-borns, children and adults in clinical environments and in everyday conditions. As part of this development, we are investigating novel, advanced light emitting diodes that we hope to incorporate into a new generation of wearable brain imaging devices. In this project, the student will be responsible for analysing and characterising these novel LEDs, both in on-bench tests and in simple measurements in human volunteers. The student will gain knowledge in fields such as neuroimaging, near-infrared spectroscopy, and opto-electronics, and will learn various useful skills such as programming, applications of microcontrollers, printed circuit board design, and will also gain significant practical experimental experience. These skills will be applicable in a wide range of future studies and careers.


Encapsulation of an implant for artificial bladder control

Supervisors: Nick Donaldson and Anne Vanhoest

Student: Madeline Lok

The Sacral Anterior Root Stimulator (SARS) is an implant that allows people with spinal cord injury (complete lesions) to empty their bladders when they want, and avoid urinary incontinence. About 5000 patients have had this device implanted in the last 35 years. The electrical components in the implant are insulated by adhesive silicone rubber which is called the encapsulant. However, the rubber which has been used is not ‘medical’ and its application is time-consuming, so expensive.The company wants to introduce a medical rubber that can be injection-moulded. For three possible rubbers, we want to choose the best and develop the encapsulation process (cleaning, injection-moulding, curing). Then we want to compare the performance of the old and new methods by electrical and mechanical tests.


Development of an impedance meter 

Supervisors: Nick Donaldson and Anne Vanhoest

Student:  Jing Xue

An impedance-measuring device is often useful in clinical situations, such as when recording biopotentials like EEG: the impedance between the electrode and the head must be low enough that the brain signals are not spoilt by noise or interference. In a recent student project, an integrated circuit was identified (AD5933) that might be the core of a portable Impedance Meter suitable for use on people. However, the device must be adapted for 4-electrode, as well as 2-electrode measurements, the required range of impedance magnitude and frequency must be obtainable, and the device must work on actual people as well as with inanimate components on the bench. Developing the device will require both circuit design and programming because the AD5933 is set up via a microcontroller. When satisfactory results are obtained from a breadboard version, a PCB and housing should be designed and made so that a usable instrument is produced at the end of the project. It should be demonstrated with a wide range of electrodes on people.


Robust and automatic registration of multi-angle all-optical ultrasound images

Supervisors: Erwin Alles and Adrien Desjardins

Student: Nina Montana Brown

All-optical ultrasound is an emerging imaging modality where ultrasound is both generated and detected using light. Previously, a bench-top imaging setup was presented that achieved video-rate 2D imaging of biological tissue samples, albeit at limited dynamic range and high artefact levels. Recently, a rotation stage was added to the imaging setup to allow for imaging of small samples from multiple angles, around an unknown and moving origin of rotation. The aim of this project is to process, register, and merge the resulting stack of images into a single image, thereby strongly improving the image quality. This project will consist primarily of signal processing and programming, and hence prior experience with programming languages (such as MATLAB or Python) is desirable.


Evaluating an open source Radiotherapy Treatment Planning System

Supervisor: Stacey Holloway and Megan Wilson

60% of cancer patients in the UK undergo radiotherapy as part of their treatment, and 40% of these patients are cured. Radiotherapy uses radiation to kill cancer cells and the greater the radiation dose, the greater the chance of cure. Planning a radiotherapy treatment begins with taking a CT of the patient, this provides both geometric information about the tumour and surrounding healthy tissues. The CT also provides information required for dose calculation in the patient. Physicists map out the radiation using a Treatment planning System (TPS) and either forward or inverse optimisation techniques to best achieve the doctor’s treatment goals. MatRad is an open source TPS written in Matlab. We are looking of a self driven student with an interest in radiation physics or medical physics and experience coding in Matlab to evaluate the potential for using MatRad as both a research and teaching tool. This project will be exclusively computational, an interest in coding is essential.


Understanding mineral structures present in the diseased human brain.

Supervisors: Sergio Bertazzo and Elena Tsolaki

Student: Stefanie Shi           

Pathological mineralisation processes are associated with a range of diseases including atherosclerosis, rheumatic fever, breast cancer and ovarian cancer, Parkinson’s and Alzheimer’s disease. Mineral deposits have been observed in many structures present in the brain. Mineralization of the basal ganglia which has been considered to be a result of aging for many years, however noted that large quantities of mineral deposits are correlated to cognition impairment and psychosis. Mineral deposits in brain tissue even though a very common phenomenon are yet not very well understood with fundamental information missing on the processes involved. Preliminary data has suggested a correlation between mineral deposits and Alzheimer’s disease. This study aims to understand the origins of one of such structures, through artificial reproduction of the structures. Animal brain samples will be used and treated with different chemical methods to characterise the induced mineral structures and evaluate their similarities/differences to the naturally occurring structures found in the diseased. This innovative research will build on the comprehensive skill set of the applicant and take it in a new and exciting direction that focuses on truly medically important breakthroughs, such as the role of brain mineralisation in Alzheimer’s disease. The student will participate in the development of the project, sample collection and preparation, experimental measurements, analysis of the data and interpretation. The project will involve lab work and use of scanning electron microscopy and optical microscopy, with new techniques potentially added as the research work progresses. This is an experimental research project where laboratory experience is a plus but not a requirement.


The role of neurofibrillary tangles in the formation of mineral deposits in the brian of Alzheimer’s disease patients.

Supervisors: Sergio Bertazzo and Elena Tsolaki

Student: Astrid Suryandari           

Pathological mineralisation processes are associated with a range of diseases including atherosclerosis, rheumatic fever, breast cancer and ovarian cancer, Parkinson’s and Alzheimer’s disease. Mineral deposits have been observed in many structures present in the brain. Mineralization of the basal ganglia which has been considered to be a result of aging for many years, however noted that large quantities of mineral deposits are correlated to cognition impairment and psychosis. Mineral deposits in brain tissue even though a very common phenomenon are yet not very well understood with fundamental information missing on the processes involved. This study aims to understand the origins of the mineral deposits, through artificial reproduction of the structures. In vitro synthesis will take place and mineralisation will be induced in order to evaluate the similarities/differences to the naturally occurring structures found in the diseased. This innovative research will build on the comprehensive skill set of the applicant and take it in a new and exciting direction that focuses on truly medically important breakthroughs, such as the role of brain mineralisation in Alzheimer’s disease. The student will participate in the development of the project, sample collection and preparation, experimental measurements, analysis of the data and interpretation. The project will involve lab work and use of transmission electron microscopy, with new techniques potentially added as the research work progresses. This is an experimental research project where laboratory experience is a plus but not a requirement.


Optimisation of platelet preparation protocol.

Supervisors: Sergio Bertazzo and Elena Tsolaki

Student: Eda Aydin            

In this project, we will use state-of-the-art electron microscopy techniques to optimise a preparation protocol for the visualisation of platelets which then will be used to better understand the role of such cells in cardiovascular diseases proposing and consequently will allow the development of new treatment methodologies. The student will participate in the development of the project, sample collection and preparation, experimental measurements, analysis of the data and interpretation. The project will involve testing different preparation methods and imaging of the samples using a transmission electron microscope to evaluate the quality of the preparation methods, new techniques might be potentially added as the research work progresses. This is an experimental research project where laboratory experience is a plus but not a requirement.


Magnetic field hyperthermia

Supervisors: Quentin Pankhurst & Paul Southern

Student: Cliona Maley

Magnetic field hyperthermia (MFH) is a phenomenon in which specific materials (we use iron oxide nanoparticles) can be made to heat up quickly (say, room temperature to 100 °C in less than a minute) by being placed in a time-varying magnetic field. It is being explored for several biomedical applications, where the ability of magnetic fields to penetrate deep into tissue means that we can remotely heat targets that are far away from the skin, such as having been injected into a cancer tumour. We are working on many different aspects of MFH in the UCL laboratories in the Royal Institution in Mayfair, focusing primarily on the underlying physics of the heat transfer, and the systems engineering of the magnetic field generators. Depending on the students skills, the project will be designed to complement existing work and generate new insight into this important new therapeutic modality.


Magnetic Field Hyperthermia - On-the-fly monitoring of thermal dose delivery


Supervisors: Quentin Pankhurst & Paul Southern

Student: Haadi Iftikhar

Magnetic field hyperthermia (MFH) is a phenomenon in which specific materials (we use iron oxide nanoparticles) can be made to heat up quickly (say, room temperature to 100 °C in less than a minute) by being placed in a time-varying magnetic field. It is being explored for several biomedical applications, where the ability of magnetic fields to penetrate deep into tissue means that we can remotely heat targets that are far away from the skin, such as having been injected into a cancer tumour. We are working on many different aspects of MFH in the UCL laboratories in the Royal Institution in Mayfair, focusing primarily on the underlying physics of the heat transfer, and the systems engineering of the magnetic field generators. Depending on the students skills, the project will be designed to complement existing work and generate new insight into this important new therapeutic modality.


Matlab GUI for efficient and flexible signal processing of Fast Neural EIT

Supervisors: Kirill Aristovich and Mayo Faulkner

EIT is a technique which can be used to image activity in the brain and nervous system using EEG type electrodes on the scalp. It is currently being investigated for real time imaging of epileptic activity in the brain, and of nerve activity for bioelectronical medicinal applications. The pipeline of the signal processing in EIT is such that it is mathematically and computationally intensive, and its current state is an assembly of different algorithms assembled together via scripting. There is current need for the growing EIT community to have a universal tool which can automate data processing and aid data quality assessment following acquisition of data from an EIT system. The purpose of the project is to create such a tool. It will include a Graphical User Interface integrated together with the existing data processing algorithms, and a quality assessment vehicle through which data can be objectively evaluated, checked, rejected, and saved for further processing. The tool will be coded using the Matlab platform and integrate the latest advanced algorithms, filtering, digital demodulation, and advanced data visualisation, which are currently used in EIT.


Neural electrode discharge following monophasic pulses

Supervisors: Henry Lancashire & Nick Donaldson

To avoid tissue and electrode damage during neural stimulation electrodes must remain within safe potential limits. In addition, stimulation is balanced by current reversal to discharge the electrode and reverse the electrochemical processes at the electrode. If, after a first negative or positive pulse, the current is not reversed and the electrode is left open circuit slow discharge takes place. This project aims to understand the processes occurring during slow discharge of stimulating electrodes.
During this project you will:
Design and test a current controlled stimulator for delivering monophasic and biphasic pulses.
Manufacture microelectrodes in a cleanroom environment.
Investigate the electrode response to stimulation pulses in a range of test solutions.

Modelling neural electrode charge injection in vitro

Supervisors: Henry Lancashire & Nick Donaldson

Student: Omar Shafi

To elicit neural responses stimulation electrodes must achieve threshold levels of charge injection. However, safe charge injection is limited by irreversible reactions with occur if electrodes are polarised beyond safe limits. If these irreversible reactions occur electrode corrosion, or worse nerve damage, results. Measurements in vitro (in saline) often overestimate clinically safe limits by as much as 10×. This means that tissue damage will occur if charge limits predicted in vitro are used with implanted electrodes. This project aims to develop an in vitro model in which charge injection response mimics the implanted environment. 
During this project you will:
Manufacture microelectrodes in a cleanroom environment.
Produce in vitro tissue models.
Electrochemically test of microelectrodes in vitro.
Analyse electrochemical data using statistical methods.


Assessment of nasal blockage with acoustic sensors

Supervisors: Terence Leung and Peter Andrews

Student: Daniel Mapana

Nasal blockage is a common condition which could indicate a range of pathologies from common cold to tumours. While a patient can describe nasal sensation, the information is often subjective, inaccurate and lacking in detail. There is also a growing need for quantifying nasal blockage severity as it is becoming increasingly important for clinicians to provide evidence for their medical/surgical interventions. The aim of this project is to develop a nasal blockage analyser to assess the degree of nose blockage. Acoustic sensors are used to measure the nasal airflow at the nostril opening through recording air turbulence sounds. The bio-acoustic signals provide a novel, accurate and yet simple way to characterize and quantify nasal airway conditions naturally and objectively. The role of the student will be to participate in analysing data collected from the Royal National Throat, Nose and Ear Hospital and analyse them. The project is especially suitable for a medical student who is interested in ENT and would like to have early clinical contacts with patients. Interested students are encouraged to get in touch with Terence sooner rather later because of the need to apply for a research passport.


Monitoring diabetic foot ulcers with 3D imaging

Supervisors: Terence Leung  and Janice Tsui

Student: Melody Langroudi

Diabetic foot ulcers are chronic wounds which require management by multidisciplinary teams across primary, secondary and tertiary care. To aid in communication and management of these wounds, there is a need to monitor the wound size accurately and quickly, taking into consideration complex shapes on uneven surfaces. We have been developing a low-cost technique to quantify wound size by filming the foot ulcer with a smartphone, followed by image analysis of the video footage using a software package known as NukeX. This software package was originally designed for the film industry to create visual effects in films such as Gravity and Star Wars. We are collaborating with the software developer, Foundry (www.foundry.com), to repurpose the software to perform 3D surface area measurement in a clinical environment.  This project will require the student to collect data from patients with diabetic foot ulcers in Royal Free Hospital. It will suit a medical student who is interested in vascular surgery, 3D modelling software, digital photography and would like to have early clinical contact with patients.


Detecting anaemia in pre-school aged children and pregnant women using smartphone photography

Supervisors: Terence Leung, Judith Meek and Sara Hillman

Student: Fiona Young

Anaemia is a condition in which the haemoglobin concentration in the blood is reduced, impairing its capacity to transport oxygen. The World Health Organisation (WHO) estimated that anaemia affects a quarter of the world’s population. Pre-school aged children (<5 years old) and pregnant women are especially susceptible to developing anaemia. Severe anaemia in pregnant women is also associated with a low birth weight of the baby, stillbirth and newborn death. In children, anaemia can compromise cognitive and physical development, which translates into a huge socio-economic burden.  Once the condition has been identified, most cases of anaemia can be treated effectively.We are developing a smartphone-based technique to diagnose anaemia based on the colour of blood vessels in the eye area. The student will have the opportunity to join our multinational team made up of clinicians and scientists from the UK, Ghana and India, and help us develop a non-invasive imaging technique for screening anaemia in pre-school aged children and pregnant women. This project will suit a student who is interested in global health, signal and image analysis, statistics and digital photography.


Design of a home monitoring wearable device for cirrhosis patients

Supervisors: Terence Leung, Prabhav Reddy and Raj Mookerjee

The heart rate of a healthy person has certain variability, meaning that the duration between successive heart beats varies slightly over time. In fact, the loss of heart rate variability (HRV) can indicate pathological conditions not necessarily related to the heart itself.It has been shown that a drop of HRV in cirrhosis patients often relates to decompensation episodes (Mookerjee Lab), which if discovered early can enable improved care management of the patient. The aim of this project is to design an HRV monitor that a patient can use at home. This home monitoring device will consist of ECG amplifiers, a microcontroller (Arduino) and a Bluetooth wireless transmitter, all packaged in a compact enclosure comfortable to wear.During the design phase of the project, the student will have the opportunity to visit Royal Free Hospital to interview cirrhosis patients and clinicians to ensure that the design satisfies the criteria of end users. The student will assist in validating algorithms developed to identify HRV patterns associated with decompensation episodes from existing HRV data and prospective collection using the new device. The final version of the prototype will go through safety tests and be trialled on cirrhosis patients.This project will suit an engineering student who is interested in electronics, programming, signal analysis, product design and medical device development.


Development of X-ray and ultrasound imaging phantoms for clinical training of minimally invasive procedures

Supervisors: Efthymios Maneas, Premal Patel, Wenfeng Xia and Adrien Desjardins

Students: Moe Kashima and Alex Lane

X-ray (computed tomography / fluoroscopy) and ultrasound imaging are widely used to guide minimally invasive procedures. Tissue-mimicking phantoms have been shown to be valuable tools for quantifying imaging performance and for training of young practitioners. This project will involve the development and evaluation of new types of x-ray and ultrasound imaging phantoms using open-source software to extract anatomical structures and generate moulds, 3D printing and novel soft tissue-mimicking materials. Potential clinical applications could include liver resection or stent placement. The students will work closely with our clinical collaborators to test and refine these phantoms.


Development of patient-specific ultrasound and photoacoustic imaging phantoms to guide minimally-invasive procedures

Supervisors: Efthymios Maneas, Jonathan Shapey, Wenfeng Xia and Adrien Desjardins

Student: Jayden Patel

Tissue-mimicking phantoms are crucial for the development of ultrasound and photoacoustic imaging systems and for clinical training. Ideally, phantoms for this modality comprise tissue-mimicking materials with tuneable optical properties and which are stable over time. This project will involve the development and evaluation of new phantoms using optical and acoustic properties measurements, imaging systems and novel soft tissue-mimicking materials. 3D printing and laser cutting techniques will be used to generate moulds of patient-specific vascular structures. Potential clinical applications could include surgical needle biopsies or catheter placement. The students will work closely with our clinical collaborators to test and refine these phantoms.


Design and Implementation of a 3D Magnetic Tracking System for Neuro-navigation

Supervisors: Hubin Zhao and Rob Cooper

Neuro-navigation is a set of computer-assisted technologies used by researchers, neuroscientists and neurosurgeons to guide or "navigate” around the brain. A common application of neuro-navigation is to use three-dimensional positioning methods to guide the location of sensors, stimulators or sensors. We are now developing a 3D positioning methodology to help guide the location of novel neuroimaging technologies to support accurate imaging of human brain function.
In this project, the student will be responsible for designing and implementing a proof-of-concept magnetic tracking system for neuro-navigation. The student will gain knowledge in fields such as neuroimaging, magnetic tracking, and power & optical electronics, and will learn various useful skills such as programming, applications of microcontrollers and printed circuit board design. The student will also gain significant practical experimental experience. These skills will be applicable in a wide range of future studies and careers.


Estimation of geodesic distances and its application to predict local features on the surface of the brain

Supervisors: Andre Altmann and Eugenio Iglesias

The brain is one of the most complex organs. Parts of the human brain, in particular the cerebral cortex, consist of layers that are folded in order to be squeezed into the available space of the skull. Modern imaging techniques such as magnetic resonance imaging (MRI) enable us to acquire structural and functional information about the entire brain in vivo although at a limited resolution. Complementary post-mortem samples are often used to gather additional information about the brain architecture that cannot (yet) be acquired using in vivo (non-invasive) imaging. Moreover, acquiring the high-resolution post-mortem data is often labor intensive and costly. Thus, these data are not available in a brain-wide fashion. One such example are gene expression measurements that are sparsely sampled across the brain (e.g., at a few hundred locations).
        In this project we aim to (i) measure distance on the brain surface (geodesic) which (ii) can be used for kernel based interpolation methods. As an intended application we will infer gene expression for any given point on the human cerebral cortex.


Respiratory motion models from unsorted 4DCT data for planning and guiding radiotherapy

Supervisors: Jamie McClelland and Bjoern Eiben

Student: Korn Pavavongsak

Four Dimensional Computed Tomography (4DCT) is often used to plan radiotherapy treatment of lung cancer and other tumours that move with respiration. It is based on the assumption that the motion due to breathing is regular and reproducible from breath-to-breath. However, this is often not the case, leading to ‘sorting artefacts’ in the 4DCT scans and errors and uncertainties in the Radiotherapy treatment.At the Centre for Medical Image Computing (CMIC) we have been developing respiratory motion models that can be fitted directly to the unsorted 4DCT data, and hence can account for breath-to-breath variations in the respiratory motion. This results in a high quality artefact free image, as well as a model that can estimate respiratory motion including the breath-to-breath variability.This project will validate and characterise the performance of the motion models using a variety of simulated and real clinical data, and investigate how the models can be used to better plan and guide radiotherapy treatment. There is the possibility that a successful project could lead to a conference or journal publication. It will require good computing and mathematical skills.


Respiratory motion modelling: exploring temporal validity

Supervisors: Bjoern Eiben, Elena H. Tran and Jamie McClelland

Motion models have the potential to play an important role in radiotherapy planning and guidance, as well as other medical imaging procedures in the thorax and abdomen. They allow estimation of the internal motion of a patient based on one or more easily acquired ‘respiratory surrogate signals’.To build the motion models, image registration is usually used to measure the internal motion in a set of training images. Thereafter a correspondence model is fitted in order to relate the measured motion with the surrogate signal. If a medical treatment is performed on the basis of a motion model, it has to be shown that the model is accurate within specified limits. The relationship between the surrogate signal and the internal motion however can change over time which in turn means that the motion model will no longer be valid and needs to be updated. This project aims to explore methods to measure the model accuracy and on this basis to evaluate for how long a motion model is valid. The student will work with open-source image registration software and cine-MR images of the lung of volunteers that were acquired over the course of about 30 minutes. This project will be suitable for a student with strong maths and computing skills. 


Origin of calcification in the cardiovascular system

Supervisors: Sergio Bertazzo and Elena Tsolaki

Student: Emma Ponting     

           There are about 7 million people in the UK suffering from cardiovascular diseases, with calcific aortic valve disease been the most common vascular disorder in economically developed countries. It has been observed that cardiovascular calcification, which is characterised by calcific lesions on valve leaflets causing impaired movement and on arteries leading to stenosis, it is actually present in 100% of the world population above 45 years old. Physicochemical analysis carried out by Dr Bertazzo has showed that the mineralised structure is a highly crystalline apatite (a form of calcium phosphate), which are believed to be originating from cells in the blood stream. In this project, we will use chemical, biological and state-of-the-art electron microscopy techniques to isolate these cells and evaluate a series of mineralizing protocols to evaluate the capacity of these cells to calcify. The student will participate in the development of the project, sample collection and preparation, experimental measurements, analysis of the data and interpretation. The project will involve testing different preparation methods and imaging of the samples using a transmission electron microscope to evaluate the quality of the preparation methods, new techniques might be potentially added as the research work progresses. This is an experimental research project where laboratory experience is a plus but not a requirement.


Novel PET-MR phantom design using materials with tissue properties for both scanning technologies

Supervisiors: Pawel Markiewicz and Daniil Nikitichev

This work addresses the pressing need for the accurate evaluation of PET/MR performance and widens understanding of the accuracy and precision of PET/MR measurements as well as the factors that affect PET quantification.  The project designs and constructs multi-modality PET/MR brain phantoms mimicking the properties of brain/head tissues for a realistic representation of the measured signal by both modalities. It will involve novel 3D printing of the skull, preparation of gels for mimicking the brain-tissue, performing experimental PET/MR scans, and finally to analyse the data using the open-source software NiftyPET. This will facilitate a comprehensive evaluation of the full image generation and processing pipelines involved in quantitative and multiple time-point (longitudinal) neuro-imaging.  The project addresses the current lack of such phantoms and methods, which would enable simultaneous evaluation of MR and PET image quality for clinical trials of new therapeutics in dementia and other diseases of the brain.


Precision PET imaging in accurate diagnosis of disease

Supervisiors: Pawel Markiewicz and Frederik Barkhof

Student: Syafiq Ramlee

The reputation of PET as a quantitative imaging tool (i.e., reconstructed images can be calibrated in absolute units of radioactivity concentration) is largely based on the fact that, among others, an exact correction for photon attenuation, scatter and random coincidences as well as robust image analyses are available. The project will investigate the effects of quantitative corrections required in PET image reconstruction and analysis using different PET radiotracers, such as 18F-FDG, 18F-FDOPA, 18F-choline, 18F-florbetapir in neuro-oncology, epilepsy, and dementia.

Quantitative PET scans rely on detecting and precise counting of photon pairs emitted back-to-back from different tissue parts while correcting for unintended measurements.  Due to significant interactions with the tissue on their trajectories, some photons are deflected (attenuated) from their straight paths causing measurement errors. Additional errors are caused by the limited speed of detection process, due to which some measured photon pairs will be random and false.  Furthermore, the noise due to limited counting statistics makes challenging the registration of MRI structural images to PET images for more informed PET quantification.

No prior programming skills are needed but it requires dedication and enthusiasm to learn Python (support and help will be provided).  The student will have the opportunity to learn (i) different PET radio-distributions in health and disease, (ii) physics of PET acquisition (iii) mathematics/statistics of image reconstruction and analysis.


Development of novel corrective methods for quantitative PET-MR neuro-imaging

Supervisiors: Pawel Markiewicz and Jem Hebden

The reputation of PET as a quantitative imaging tool (i.e., reconstructed images can be calibrated in absolute units of radioactivity concentration) is largely based on the fact that, among others, an exact correction for the measured random and scatter photon pairs is achievable.  The ability to accurately and precisely represent the radiotracer concentration in the body is important in order to ensure that the PET images reflect the underlying physiology in health or disease with high accuracy, precision, and spatial resolution.

Random coincidences are recorded as accidental detection of two unrelated photons within the measurement time (coincidence time window). Due to the limited speed of the detectors and the time of flight (TOF) of emitted gamma rays, the coincidence window also allows recording of random events.  For quantitative imaging, the random coincidences have to be accurately estimated and accounted for. Currently, the random coincidences are estimated either by a separate measurement, while the purpose of this work is to provide an estimate of the random data without any additional measurement.  This is achieved by noting that the distribution of random data is smooth and identifying regions in the projection data which are random only, based on which the overall random photon pairs are estimated.

One of the other key aspects in high accuracy quantitative image reconstruction is correction for scatter events. The novel approach of this project is to build a unified fully 3D time of flight (TOF) scatter modelling which inherently accounts for the difference in detection technologies (TOF/non-TOF) as well as the large field of view of PET/MR scanners. The key aspect of the proposed model is the voxel-driven scatter modelling, where each emission voxel is treated independently, for each of which a separate 3D probability scatter distribution is found.

No prior programming skills are needed but it requires dedication and enthusiasm to learn Python (support and help will be provided).  The student will have the opportunity to learn (i) different PET radio-distributions in health and disease, (ii) physics of PET acquisition (iii) mathematics/statistics of image reconstruction and analysis.


High throughput kinetic analysis of dynamic PET acquisitions for more precision disease progression

Supervisiors: Pawel Markiewicz and Kjell Erlandsson

Student: Rohan Misra

The non-invasive methods of measuring amyloid plaques in the brain using positron emission tomography (PET) has widen the understanding of the molecular pathology of dementia.  PET as a fully quantitative imaging tool, has the ability to accurately and precisely map the radiotracer concentration in the body.  This is particularly useful in dynamic scanning followed by kinetic modelling of the time-varying distribution of any amyloid radiotracer in the body and thus quantify additional physiological parameters of interest, such as the binding potential, otherwise unavailable with static scanning.

This project aims at designing high-throughput kinetic analysis of PET amyloid image data from very large datasets procured from various dementia centres. The datasets will have to be first pre-processed and normalised so that they can be pooled for the unified kinetic analysis.  The processing and understanding will be performed using the NiftyPET software platform, for which full support will be given.

No prior programming skills are needed but it requires dedication and enthusiasm to learn Python (support and help will be provided).  The student will have the opportunity to learn (i) PET radio-distributions in dementia, (ii) physics of PET acquisition (iii) mathematics/statistics of image reconstruction and analysis.


PET-MR phantom harmonization study across seven Dementias Platform UK centres.

Supervisiors: Pawel Markiewicz and Frederik Barkhof

Phantom measurements are important for quantifying the performance of imaging instruments and that they meet manufacturer specifications. This helps understanding how the PET/MR technology with n all the major aspects of applied physics and statistics can help diagnose disease in clinical/health environments. 

Knowledge of the true underlying distribution in phantom measurements enables the assessment of bias, while repeated measurements enable the measurement of precision of PET/MR imaging systems. However, although phantom measurements provide a general guide, they are unlikely to be an accurate representation of clinical imaging in practice. The PET phantom experiments are used primarily for characterisation of the PET component of PET/MR scanners across multiple sites and consisting of the Siemens Biograph mMR and the GE Signa scanners.  Secondarily, based on the phantom scans, it is endeavoured to propose a PET phantom imaging protocols for qualifying PET/MR scanners for clinical trials.

No prior programming skills are needed but it requires dedication and enthusiasm to learn Python (support and help will be provided).  The student will have the opportunity to learn about different PET/MR scanners and centres in the UK with a possible visit to one of the centres.


Determine the geometric uncertainties of a proton therapy treatment

Supervisor: Gary Royle
Proton therapy is a relatively new form of cancer treatment which can potentially provide very precise targeting of the tumour. UCL Hospital is currently building a proton therapy centre to treat deep tumours in adults and children. Currently there are some uncertainties in precisely where the radiation dose is delivered within the patient. Some are related to the physics and some to the biology. This project will focus on exploring the physical uncertainties related to the geometry of the beam and of the patient. 
Aim - quantify the dose to tissue immediately surrounding the treatment volume in proton therapy.
Tasks 
 - identify the various uncertainties in the precise location of the proton dose spot within the patient, both distally and laterally.
 - quantify the identified uncertainties for a simple beam arrangement in a simple, homogeneous tissue.
 - calculate total uncertainty and correlate with reported literature to validate.
 - calculate for more complex beam arrangements in a simple, homogeneous tissue.
 - calculate for complex beam arrangements in a patient. 
Outcomes
 - an itemised method for quantifying proton beam geometrical uncertainties.


Improving the accuracy of radiotherapy to lung tumours
Supervisor: Gary Royle
Lung cancer is the most common form of cancer worldwide, with poor survival rates for patients. A range of treatment options exist for the various types of lung tumour, including surgery, radiotherapy and drug treatment. This project will focus on the accuracy of radiotherapy. All tissues within the body close to the diaphragm move continuously. This causes a problem for radiotherapy treatment of tumours within this region. Current practice is varied, ranging from treating a large volume across the movement envelope to controlling the breathing of patients to monitoring breathing patterns and modifying treatment delivery. This project will cover all three options and focus on the latter. 
Aim - study the correlation between internal tumour motion and external markers
Tasks
 - understand internal and external breathing patterns.
 - review the literature correlating various surface surrogate markers.
 - apply to the different breathing modes.
 - quantify effect to patient of potential errors of surface markers.
 - estimate radiation dose effect to patient with and without breathing mitigation techniques.
Outcomes
 - a study of the potential errors associated with breathing mitigation for lung cancer patients undergoing radiotherapy.


Patient localisation in precision radiotherapy treatments
Supervisor: Gary Royle
Precision radiotherapy is guided by medical imaging to deliver a targeted beam of radiation at the tumour volume. Whilst the instrumentation can be very precise it requires the patient not to move during the course of treatment, which can be minutes. For some body regions it is critical that the patient does not move, to avoid hitting highly radio-sensitive organs. Commonly plastic shells are made to conform to the patient’s anatomy and hold them in position. These work sufficiently in some cases but less well in others. This project will explore a more accurate alternative.
Aim - to generater precise surface maps of patients for 3D printing radiotherapy shells
Tasks
 - understand clinical requirements of radiotherapy shells 
 - review the literature to determine the current options and their advantages and disadvantages
 - develop a method using surface scanning technology
 - evaluate performance on simple geometric shapes
 - translate to complex anatomical shapes
 - convert to a 3D printing file format
Outcomes
 - a method for generating a 3D printable map of an anatomical shell


Modelling the risk of second cancers in paediatric neuroblastoma treatment

Supervisors: Catarina Veiga

Student: Ammar Alhadi

Childhood cancer treatment has become relatively successful, and nowadays 75% of the patients survive for 10 years or more. Radiotherapy (RT) had a fundamental role in improving outcomes, but with increasing survival the late treatment-induced side-effects also become more important. Namely, ~20% of survivors are expected to develop new primary cancers decades later which were induced by the RT they received during their childhood. In this project the student will investigate the risk of second cancers in a cohort of young children treated for neuroblastoma, a rare type of cancer that usually occurs in the abdominal region. The student will tailor existing models of risk to this particular cohort, and compare different RT modalities (i.e., conventional RT, intensity modulated-RT and proton therapy). The project will suit a student that enjoys programming and mathematical modelling. Matlab knowledge is desirable.


Calculating acoustic radiation forces for particle manipulation

Supervisors: Ben Cox and Bradley Treeby

As well as being useful for imaging, ultrasound can be used to impart forces on objects. In recent years there have been demonstrations of various trapping and manipulation devices using acoustics, including acoustic tweezers and sonic screwdrivers. This project will be a theoretical, mathematical and numerical study of the radiation forces that can be achieved with conventional piezoelectric and laser-generated ultrasound sources. The student will be expected to study the fundamental physics behind acoustic radiation forces so that they are able to write algorithms in Matlab that interface with the acoustic modelling software k-Wave (www.k-wave.org) to calculate the radiation forces from arbitrary acoustic fields.  As well as calculating the forces for single-frequency wavefields, eg. using Gork’ov potentials, the student will explore the extent to which pulsed, broadband, acoustic fields can be used to manipulate objects. They will therefore will need a good grasp of physics, mathematics of partial differential equations, and scientific computing, and a high degree of motivation. All students working with the Biomedical Ultrasound Group (BUG) are expected to spend at least one day per week, or two half days, working in the group in MPEB. Additional work outside these hours will also be necessary. All students are expected to attend the weekly BUG meeting, and are expected to present their latest results to the rest of the group once every month.


Parameter optimisation for x-ray phase contrast CT of excised breast tissue 

Supervisors: Charlotte Hagen and Lorenzo Massimi

X-ray phase contrast computed tomography (XPC-CT) stands for a class of 3D radiographic imaging techniques, which, in addition to x-ray attenuation, are sensitive to phase shifts. This is especially beneficial for the imaging of weakly attenuating biological tissues.

XPC-CT techniques are investigated by an increasing number of groups worldwide, including the Advanced X-Ray Imaging (AXIM) group at UCL.

AXIM currently investigates the use of XPC-CT for in-theatre scanning of breast tissue specimens excised during surgery, and aims to demonstrate that it provides improved delineation between tumour and healthy tissue. The student on this project will work alongside the AXIM team and explore the effect of different scan parameters, with the aim of determining those which maximize contrast between tumour and healthy tissue.

The project will be based on simulations (a simulation code implemented in Matlab will be provided), with a possibility of verifying results experimentally. 

Skills required: knowledge of medical imaging with ionising radiation, image quality metrics, image analysis, programming in MATLAB.


Understanding X-ray scattering: a new contrast mechanism for medical imaging

Supervisors: Peter Munro and Fabio Vittoria

The absorption and refraction of X-rays is now well understood and applied in the clinic. X-rays are, however, also scattered, due to tissue having features which are smaller than the resolution of the imaging system. The X-ray scattering signal is thus a measure of statistical properties of tissue. We have a research program currently focussed on this and have projects available in a range of areas including theory, development of computational models, implementing algorithms on GPUs, phantom construction and tissue imaging. This project can thus be adapted to the interest of the student. Feel free to come and chat with us if you’re interested in working on the development of a new imaging modality. Good knowledge of wave optics is important for this project.


Making phantoms for elastography

Supervisors: Peter Munro and Alessandro Olivo

A project is underway within the advanced X-ray imaging group to perform elastography using X-ray phase imaging. Elastrography obtains images of the mechanical properties of tissue and is performed by imaging how tissue deforms in response to a mechanical load. Part of this project requires the development of phantoms with controllable mechanical properties. This project entails the construction of such phantoms along with their characterisation using an instron machine which provides gold standard characterisation of the mechanical properties of materials. This project is largely experimental, though an elementary understanding of continuum mechanics will be beneficial.


Signal processing in optical coherence tomography

Supervisors: Peter Munro and Callum Macdonald

Optical coherence tomography (OCT) has developed from lab based first principle demonstration to routine clinical tool in ophthalmology in just 25 years. OCT Images are generated from measured signals using reasonably simple and reliable techniques, namely, discrete Fourier transforms. There still exists some uncertainty about how traditional sampling theory applies to OCT image reconstruction. This has important ramifications for how OCT images are analysed. This project has a strong mathematical and computational focus, however, it has the potential to impact a variety of functional imaging approaches based on OCT. Good knowledge of Fourier transforms is important for this project.


Low-cost ultrasound array demonstrator

Supervisors: Ben Cox, Bradley Treeby and  Panayiotis Georgiou

Diagnostic ultrasound images are formed line-by-line from an array of ultrasound source-receivers. In this project, the student will build a giant-sized version of a diagnostic ultrasound machine from ultrasonic parking sensors that will work in air. It will be designed for use in outreach and demonstration, to show the principles of ultrasound imaging. The student will be required to design and 3D printing the housing for the transducers, design and construct the circuits connecting them to a multiplexer controlled by a Raspberry Pi. They will also write the software for controlling the array and forming and displaying the image. The student will need a good grasp of physics, electronics and programming of microprocessors and a high degree of motivation. All students working with the Biomedical Ultrasound Group (BUG) are expected to spend at least one day per week, or two half days, working in the group in MPEB. Additional work outside these hours will also be necessary. All students are expected to attend the weekly BUG meeting, and are expected to present their latest results to the rest of the group once every month.


Optical Calibration using Deep Learning

Supervisors: Bongjin Koo, Matt Clarkson

    Laparoscopic (keyhole) surgery has well-known benefits for the patient including reduced physical trauma, fewer post-operative complications and faster recovery time. However, in liver surgery, only 10% of surgery is performed laparoscopically, the remaining 90% is performed as open-surgery. Laparoscopic surgery is more complicated and difficult to perform than open surgery, so when a surgeon decides that the risk of performing a procedure laparoscopically is too high, then they choose open surgery. The SmartLiver project at UCL has been developing an augmented reality image-guidance platform, to provide information from pre-operative CT data to the surgeon, and overlay it in real-time on the intra-operative, laparoscopic video image. This aims to inform the surgeon about critical structures, and hence reduce the risk of certain procedures, such that these procedures become more amenable to the laparoscopic approach. 
    The SmartLiver system uses computer vision techniques to reconstruct the liver surface, and match it to the pre-operative CT data. In order to do this the optics contained within the laparoscope must be accurately calibrated. Most current techniques rely on showing the laparoscope multiple pictures of a 2D calibration pattern. Recent advances in machine learning techniques have led to a wide array of new algorithms that outperform traditional explicitly programmed algorithms. It may be possible to develop a new calibration method, based on machine learning, that is more robust and requires fewer views than is currently necessary and yet leads to results that are comparable to, or even better than traditional methods. This project will investigate the following
    A) How to generate a large number of training samples by rendering views of a calibration pattern
    B) Implementation of a network using TensorFlow
    C) Training a network based on image or geometrical features
    D) Evaluating the predictive capability of the calibration algorithm in our mock operating room with a group of volunteers
Pre-requisites: Willingness to learn TensorFlow, ability to program in Python, basic data analysis, understanding of some geometry, good experimental rigour


Optical Calibration using 3D calibration targets.
Supervisors: Matt Clarkson, Daniil Nikitichev
    Laparoscopic (keyhole) surgery has well-known benefits for the patient including reduced physical trauma, fewer post-operative complications and faster recovery time. However, in liver surgery, only 10% of surgery is performed laparoscopically, the remaining 90% is performed as open-surgery. Laparoscopic surgery is more complicated and difficult to perform than open surgery, so when a surgeon decides that the risk of performing a procedure laparoscopically is too high, then they choose open surgery. The SmartLiver project at UCL has been developing an augmented reality image-guidance platform, to provide information from pre-operative CT data to the surgeon, and overlay it in real-time on the intra-operative, laparoscopic video image. This aims to inform the surgeon about critical structures, and hence reduce the risk of certain procedures, such that these procedures become more amenable to the laparoscopic approach. The SmartLiver system uses computer vision techniques to reconstruct the liver surface, and match it to the pre-operative CT data. In order to do this the optics contained within the laparoscope must be accurately calibrated. Most current techniques rely on showing the laparoscope multiple pictures of a 2D calibration pattern. Within the computer vision literature, there are methods to obtain calibration parameters using a 3D calibration target, but these methods have not been applied and evaluated in laparoscopy. Furthermore, the issue of manufacturing such a 3D calibration object in medical grade, sterilisable materials remains uncertain. This project will investigate the following

    A) Comparison of 2D and 3D calibration targets for stability and accuracy in estimating intrinsic and stereo parameters (most software provided)
    B) Construction of a viable 3D calibration target, including designing software (MATLAB) to correctly etch a pattern.

  Pre-requisites: Basic data analysis, understanding of some geometry, good experimental rigour, some programming e.g. MATLAB


Ultrasound Computed Tomography Using Two Diagnostic Linear Arrays

Supervisors: Bradley Treeby and Ben Cox

Student: Preena Patel

Diagnostic ultrasound imaging is very widely used in clinical medicine. It is low cost, portable, and generates images with high spatial and temporal resolution. However, the images are qualitatively related to the echogenicity of the tissue, and not quantitatively related to the underlying material properties, such as sound speed or absorption, which might be more closely correlated with markers of disease. Ultrasound computed tomography (USCT) is a rapidly emerging alternative to conventional ultrasound imaging that uses information about the transmission of ultrasound through the tissue in additional to reflections. In USCT, typically a ring of ultrasound elements is arranged surrounding the tissue of interest. Each element acts as an emitter one-by-one, with the other elements used to detect the ultrasound signals. The recorded data can then be used to reconstruct a quantitative image of the sound speed and acoustic absorption within the tissue. However, creating a custom array of ultrasound elements can be expensive. In this project, the suitability of using two clinical linear arrays (normally used for conventional imaging) for USCT will be investigated. An experimental rig will be created with the two arrays facing one another, with a computer controlled rotating sample holder in the centre. Custom MATLAB software will be written using the Verasonics platform to generate and detect the ultrasound waves, and control the rotation stage. Finally, different phantom objects and tissue samples will be imaged to demonstrate the accuracy and generality of the approach. All students working with the Biomedical Ultrasound Group (BUG) are expected to spend at least one day per week, or two half days, working in the group in MPEB. Additional work outside these hours will also be necessary. All students are expected to attend the weekly BUG meeting, and are expected to present their latest results to the rest of the group once every month.


Experimental characterisation of a low-frequency transducer for transcranial ultrasound neurostimulation

Supervisors: Bradley Treeby, Elly Martin, Ben Cox

Student: Lucia Albelda Gimeno

Project Overview: Transcranial ultrasound neurostimulation is a rapidly emerging therapeutic technique for brain stimulation in which low frequency ultrasound is non-invasively delivered through the intact skull. In this project, a novel ultrasound transducer designed for neurostimulation will be experimentally characterised. Free-field measurements will be made under a range of driving conditions, and the focal parameters compared with those predicted by simulations. The transmission loss when propagating through the skull will also be experimentally characterised. Finally, a coupling pad for attaching the transducer to the head will be designed and manufactured. All students working with the Biomedical Ultrasound Group (BUG) are expected to spend at least one day per week, or two half days, working in the group in MPEB. Additional work outside these hours will also be necessary. All students are expected to attend the weekly BUG meeting, and are expected to present their latest results to the rest of the group once every month.


Multi-Modality imaging phantom project:

Supervisor: Daniil Nikitichev and Jonathan Shapey

Student: Steffi Mendes

The goal of the project to develop patient-specific multi-modality (ultrasound, CT and MRI) imaging phantom to establish novel clinical procedure for brain surgery.

Using advanced 3D printing technique we will fabricate anatomical models of normal brain, cranial nerves, blood vessels, cerebrospinal fluid, skull and brain tumours based on CT and MRI data.

Over 8000 patients undergo brain tumour surgery every year in the UK. The phantom will be used to develop and validate a novel intraoperative imaging system that will improve the precision and accuracy of surgery enabling improved maximal safe tumour resection.  Further, the anatomical phantom model will be used to train surgeons to perform these complex surgeries.


Evaluating the possibility of 3D printing silicone rubber for generating tissue-like optical
phantoms.
Supervisors: Daniil Nikitichev, Rob Cooper and Jem Hebden
The UCL Biomedical Optics Research Laboratory (BORL) has extensive expertise in the development of physical models of human tissue (known as “phantoms”) which have optical properties matched to those of real human organs. These are used to evaluate new optical techniques and instruments for diagnostic monitoring and imaging, and of the brain in particular. New 3D printing technology is enabling highly sophisticated objects to be printed, including those with anatomically-realistic structures derived directly from medical images (e.g. MRI scans). The objective of this project is to investigate whether it might be feasible to adapt an existing 3D printer to generate objects from silicone rubber. Translucent silicone rubbers are available to which substances can be added to mimic the scatter and absorption of biological tissues. However, the curing (hardening) times of such rubbers are typically quite long (&gt; 30 minutes). The student will first perform an online search to explore what mechanisms have already been devised for 3D printing silicone rubber. Then the student will investigate means of accelerating the cure (e.g. through heating), and of achieving a steady and controlled production of very small volumes of cured rubber suitable for 3D printing.


Design and implementation of an optical scanning system
Supervisors: Rehman Ansari and Paul Beard

Student: Rongyu Lin
Photoacoustic (PA) tomography is a non-invasive imaging technique which uses short duration laser pulses to illuminate the tissue and excite chromophores such as haemoglobin, lipid, etc., which leads to emission of ultrasonic waves. These ultrasonic waves are recorded to produce 3D images of the absorbing structures. Miniature PA imaging probes has many clinical applications, such as in situ assessment and diagnosis of cancerous lesions in the abdominal cavity, characterization of atherosclerotic plaques in the coronary arteries, and guiding various minimally invasive surgical procedures. Our group has developed miniature forward-viewing PA imaging probes using coherent fibre-optic bundle and planar Fabry-Perot ultrasound sensor. These probes have been demonstrated to provide excellent images of 3D vasculature. The goal of this project is to extend the lateral field of view of the probe by using a curved FP sensor. This involves design and implementation of a lens scanning system that can interrogate a curved FP sensor, and evaluation of its overall performance. This project will give student a chance to learn the lens design process using Zemax software, gain experience through experimental work in optics laboratory and the opportunity to contribute to the development of a promising new miniature PA imaging probes.


Characterisation of optical ultrasound sensors for photoacoustic imaging
Supervisors: Jamie Guggenheim and Paul Beard
The aim of this project is to experimentally characterise a range of new optical ultrasound sensors developed for photoacoustic imaging. Photoacoustic imaging is a new medical imaging technique in which laser light is used to generate ultrasound waves inside biological tissue
(http://www.medphys.ucl.ac.uk/research/mle/). It provides high quality tissue images but penetration depth is limited by the sensitivity of conventional detectors. To address this, the UCL photoacoustic imaging group are developing a new class of ultrasound sensors based on optical microresonators (microscale structures that trap light). These highly sensitive devices could provide an alternative to traditional detectors that enable photoacoustic imaging deeper in tissue, paving the way to important new applications such as the detection of breast cancer
and cardiovascular disease. In this project, the aim will be to experimentally characterise a range of recently fabricated optical microresonator ultrasound sensors, this involve using existing experimental setups and software to analyse sensor properties (e.g. physical geometry) and performance (e.g. optical sensitivity), allowing investigations into the relationships between the two. This will aid the fundamental understanding of the sensors and enable optimisation of their sensitivity, contributing to the development of deep-tissue photoacoustic imaging systems.
The student will ideally have some knowledge of:
- Matlab
- Optics (e.g. lasers, lenses)
- Ultrasonics
- Experimentation


Characterisation of optical ultrasound sensors for photoacoustic imaging
Supervisors: Jamie Guggenheim and Peter Munro

Student: Mohammed Himaayat Chowdhury

The aim of this project is to use a recently developed computational model to optimise the design of optical fibre based Fabry-Pérot ultrasound sensors used in medical imaging.
Fabry-Pérot ultrasound sensors are devices that exploit optical interometry to detect ultrasound. A sensor comprises an optically clear plastic film between two highly reflective mirrors. This forms an interferometer, the thickness of which is monitored by a laser beam focussed upon its surface. Ultrasound waves induce thickness modulations in the sensor which are detected by the laser with very high sensitivity. A major advantage of these sensors is that they can be made very small whilst retaining high sensitivity. This enables miniaturised (≈100 µm diameter) sensors to be fabricated on the tips of single mode optical fibres. Such optical fibre based sensors have been developed for more than a decade at UCL and have enabled a range of new ultrasonic imaging applications such as monitoring minimally invasive heart surgery using a sensor embedded in a surgical needle. Despite their broad usage in research, it has thus far proved challenging to predict the optical performance (thus ultrasonic sensitivity) of optical fibre sensors. It has therefore been challenging to optimally design them (e.g. select optimal mirror reflectivity, sensor thickness
and material properties) to achieve the highest possible sensitivity. The aim of this project is to apply a recently developed optical model to do just this – to investigate the impact of several design parameters thereby providing insights into the development of optimised sensors, improving their performance in medical imaging.
The student will ideally have some knowledge of:
- Matlab
- Optics (e.g. lasers, lenses)
- Ultrasonics
- Simulation


A Deep Convolutional Neural Network for the prediction of segmentationquality.
Superviors: Michela Antonelli and Jem Hebden
Multi-parametric magnetic resonance imaging (mpMRI) has been shown to be very effective in detecting prostate cancer and in monitoring its treatment. Since interpreting prostate mpMRI images requires a high level of expertise and is time consuming, there has been increasing interest in the development of computer-aided diagnosis systems (CAD) aimed at helping radiologists in their diagnosis and treatment of prostate cancer. Usually, the first step of a CAD system is the extraction of the anatomical region of interest from the images. The prostate consists mainly of two anatomical zones: the peripheral (PZ) and the transition zones (TZ). Cancers behave differently for different zones they originated in, thus, segmenting and characterising these regions is paramount. Several algorithms have been proposed to segment the prostate but none of them provide any information on the quality of the automatic segmentation. Since the segmentation step is crucial to the success of the following stages of lesion detection and classification, knowing how much the automatic contours can be trusted is highly desirable. The aim of this project is to develop and train a deep convolutional neural network (DCNN) that provides a quality control of image segmentation. In particular, the network will have as input a T2-weighted MRI of the prostate, an automatic segmentation, and will produce as output a number that reflects the quality of the segmentation. The DCNN will be developed by modifying an existing deep-learning software platform (NiftyNet). Coding knowledge in Python and an interest in machine learning is desirable.


Monte Carlo modelling of novel datatypes in time-domain optical imaging of tissue

Supervisors: Adam Gibson and Jem Hebden

Time-resolved optical imaging produces complex data which can be viewed and analysed in numerous ways. However, for subsequent analysis and image reconstruction, the complex data is normally simplified into datatypes. Most usually, these are intensity and mean photon flight time. Summarising the data into datatypes loses information, and in this project you will investigate a range of different potential datatypes using a Monte Carlo simulation programme. These different datatypes will be assessed for robustness, reliability and information content and a recommendation for the optimal datatypes will be made. This project is a computer simulation and will require use of complex programmes and computer programming in Matlab. Previous experience of Matlab would be valuable but is not necessary. 


 Pre-processing of time-resolved optical data

Supervisors: Adam Gibson and Jem Hebden

Data acquired using a time-resolved optical imaging system may be contaminated by patient movement, stray room light, low photon counts or instrumentation errors. Such contaminated data needs to be removed from the dataset prior to image reconstruction. The latest generation imaging system generates thousands of measurements per minute so we require programmes that can review data and automatically reject contaminated measurements. In this project, you will develop a deep understanding of measurement noise and errors and propose, develop and test algorithms that can reject bad data points. The influence of rejected data on the final reconstructed images will be assessed. This is a theoretical, computer-based project that will require computer programming in Matlab. Previous experience of Matlab would be valuable but is not necessary. 


Laser Generated Focused Ultrasound for Therapeutic Applications

Supervisors: Esra Aytac Kipergil and Adrien Desjardins

Student: Janvi Karia

Laser-generated focused ultrasound transducers involve photoacoustic generation of ultrasound from engineered surfaces. The inherent high-frequency characteristics of the generated ultrasound pulses enable acoustic waves to be tightly focused, which holds promise for the high-precision ultrasound therapies. This project will involve analysing how the generated acoustic waves depend on the transducer geometry and the spatial pattern of laser light on the transducer surface. The project will involve a combination of simulation and experiment, and close collaboration with clinical collaborators.


Investigating the effect of printing parameters on the mechanical properties of 3D printing materials

Supervisors: Eve Hatten and Rebecca Yerworth

3D printing is a manufacturing technique that is growing in popularity within Biomedical Engineering, but little has been done to fully categorise the mechanical properties of printed structures. In this project, a database will be designed and constructed to quantify properties of such structures when different print parameters are used. Areas of investigation could include: material, infill shape and density, printing temperature, precision and orientation. The student will design and print the structures, test their properties and construct a database of results. Mechanical properties of the structures will be tested using an Instron Dynamic Testing machine. Knowledge of 3D printing and mechanical testing would be an advantage in this project.


Analysis of Dance Movement from Video Using Machine Learning

Supervisors: Terence Leung and Duygu Ceylan

Student: Marina Melero

For my final year research project, I plan to work on developing a tool that understands dance movement from video and is able to segment a video from a dance routine into the specific movements it comprises of. The approach I aim to follow combines machine learning techniques with heuristics to analyze the kinematics of human motion in dance, focusing particularly on ballet. To continue with, I will explore the applications of this movement analysis method in Biomedical Engineering, especially evaluating its potential to improve current rehabilitation therapies; as a creative tool for dancers and choreographers; to improve educational resources in the performing arts; and as an automatic video editing tool for dance videos. 

This project is a continuation of the research project I carried out during my internship at Adobe Research (June-Sept 2018), where I created a method for automatic segmentation of ballet videos using audiovisual cues. My goals for the following academic year are to firstly, make this method more robust by exploring the use of deep learning models. Additionally, I plan to create a standalone application that performs this automatic segmentation and evaluate its potential opportunities in the applications mentioned above.


Evaluating factors that determine the successful development of innovations for Global Health
Supervisors: Clara B. Aranda-Jan and Jem Hebden
During the past decade, more than a thousand different innovations for global health in RLSs have been developed. However, the rate of effective designs is very limited. Most innovations hardly ever reach the ‘last mile’ of product development—failing to reach the final beneficiaries. Many of the causes for the lack of success are inherent to the product development process. Amongst some the cited reasons for failure are the lack of innovation ecosystems; the poor transferability of technologies; and overall costly development of the technologies. This research aims to understand the factors determining the successful development of innovations for Global Health. The proposed project includes the update and development of a database of innovations for global health, and the collection of data on innovations for Global Health. The skills to be developed include literature review, secondary data collection, quantitative analysis and key informant interviews.


Investigating innovations for adult Traumatic Brain Injury in the Global Health context
Supervisors: Clara B. Aranda-Jan and Jem Hebden
Traumatic Brain Injury (TBI) affects more than 50 million people each year, mostly in developing countries which account for 89% of the mortality rate. Access to surgical and rehabilitation services is key for reducing mortality related to TBI; however, developing countries account for 6% of the 313 million surgical procedures performed each year globally. State-of-the-art technologies for TBI are lacking in most RLSs, creating an urgent need for context-specific innovations to reduce fatalities. This research project proposes to investigate whether technological innovations exist that could support healthcare systems in developing countries to provide better care to TBI sufferers, improve their quality of life and, overall, reduce incidence of the condition. The investigation will focus on identifying and critically evaluating technologies that have purposefully been designed for Global Health and technologies that could potentially be transferred to/from different contexts. The skills to be developed include conducting literature reviews, secondary data analysis, qualitative methods for primary data collection, and the design of mixed research methods.


Using the Amazon Echo to monitor pediatric proton therapy patients

Supervisor: Jamie McClelland and Sarah Gulliford

Student: Chris Jackson 

Proton therapy will be used in combination with surgery and chemotherapy to treat paediatric malignancies in the UK.  Although the radiation dose distribution is optimised to minimise the dose to surrounding healthy tissues, some dose is unavoidable.  By recording the patient's experience of side effects it is possible to learn the relationship between dose to healthy tissues and toxicity.  It is then possible to optimise dose distributions which are likely to result in lower incidences of toxicity.

The Amazon Echo is the best selling voice-assistant with an estimated 27 million households adopting the devices. As well as being a popular device for entertainment and in-home support, the Echo also possess the opportunity for clinical data collection to support decision making. To date only a handful of studies have used the Echo to support patients, and only one study, in dementia, has actively collected data from patients.

 We are looking for a student to develop a prototype Amazon Echo skill (application), to interact with users and record the answers to a series of questions. In particular, the student would use the Alexa Skill Development Kit to create a lightweight skill, using Python, to ask simple questions, such as ‘how was the weather today?’, trial the skill in a group of patient advocates and use existing software to extract the answers to these questions. Appropriate methods for analysing the results will need to be investigated and implemented. We expect the work in the project, and feedback from patient advocates, to be the foundations for the development of a more comprehensive skill used to collect data from patients relating to their treatment.


An investigation into the design and feasibility of a device facilitating surgery in microgravity
Supervisors: Dr Kevin Fong, Dr Richard Colchester

Student: Eleonor Frost

There has been a continuous human presence in space for the last 19 years, an incredible achievement in itself, however this has been on the International Space Station which stays within low earth orbit (a relatively small distance). Due to this proximity, astronauts rely on Mission control for advice and most of the decisions when it comes to crew health. There has so far been no need for in-flight surgical intervention because in the event of an emergency evacuation can take as little as 3 hours. As we look at longer distance flights such as travel to the Moon and then Mars, astronauts must be able to have more independence due to the distance and communication delays. The proposition of a device which will make surgery in microgravity possible will provide this independence should an emergency surgical intervention be needed since for the first time it will be possible to perform it. There are currently no tools or protocols for surgery in space but as we go further this gap will have to be filled. The space and particularly microgravity environments presents many challenges to a surgical procedure, the main one being the different behaviour of fluids which pose a few risks if uncontrolled. This project aims to research and design a device/environment which can be used to make surgery possible and so the option of intervention isn’t ruled out and astronaut health is maintained a priority.



Developing an Unwrapping Technique for MRI Phase Imaging Using Machine Learning

Supervisors: Barbara Dymerska and Karin Shmueli

Student: Qiang Liu

The signal in Magnetic Resonance Imaging (MRI) has two constituents: magnitude and phase. In most clinical MRI studies, only the magnitude information is used. Recently, there has been rapid development of MRI methods which utilise the phase, as it can be used to increase image contrast, correct image distortions or provide information about the tissue magnetic properties (e.g. in quantitative susceptibility mapping). Phase image processing is not trivial as the range of phase values which can be measured in MRI is limited to 2π radians. This means that any phase value larger than 2π is wrapped-back into this range and phase unwrapping methods are necessary to recover the true phase values. Existing unwrapping methods are often time consuming and fail in regions which are noisy or have strong phase variations. The initial aim of this project is to develop realistic numerical models of phase variations in the human brain. These models will then be used to train a convolutional neural network to give a fast and robust solution to the phase unwrapping problem and compare it against existing unwrapping algorithms. This project will give you an insight into MRI signal properties, MRI image processing, Matlab programming and machine learning algorithms.


MRI Magnetic Susceptibility Mapping in Post-Mortem Neonates

Supervisors: Karin Shmueli and Russell Murdoch

Susceptibility Mapping (SM) is an emerging MRI technique used to characterize magnetic properties of tissues, giving insights into iron concentration, calcification and oxygenation of different tissues. Currently, the technique is applied predominantly within the brain, as the presence of large air volumes and respiratory motion during image acquisition limit its utility in the body.

 Whole-body imaging of post-mortem neonates, where these limiting factors are absent, is a potential application of SM outside the brain. Post-mortem imaging is a developing tool in assessing fetal death and the use of SM may provide information that is clinically relevant to autopsies, such as brain and hepatic iron overload. Applying SM throughout the body remains challenging, as the majority of existing methods have been designed for application in the brain. Additional challenges include small structure sizes and lack of knowledge of post-mortem neonatal tissue composition and normal MRI appearance.

The aim of this project is to investigate the application of MRI SM in post-mortem neonatal imaging. The student will learn and adopt SM techniques developed for neuroimaging and optimise them for whole body MRI of neonates. This project will give the student an insight into MRI physics, advanced image processing techniques and Matlab programming.


Developing a rapid energy dispersive X-ray imaging system for resected tissue analysis
Supervisor: JC Khong and Robert Moss

Breast cancer is the most common cancer in the UK, according to Cancer Research UK, and it accounted for 15% of all new cancer cases in 2015. Surgical removal of the cancerous tissue from the patient is among the highest primary treatment accounting for 81% of patients. A histopathological analysis will be carried out on the tumour tissue to ensure that all of the malignancy has been removed, unfortunately a second surgical procedure will be required if cancerous tissue is found at the margins (edges) of the resected tissue. To avoid secondary surgery, a rapid, automated analysis on of the excised tissue in the treatment room could be a solution. However, this has to account for the extension of the surgery time, which will increase the risks to the patient. The small differences in the X-ray attenuating properties of cancerous and healthy tissue will lead to small differences in the grey levels if solely rely on the conventional X-ray absorption imaging technique. However, with an energy-resolved pixelated detector, it is possible to combine different energy spectral to enhance the contrast. The student will be required to design phantoms to mimic the cancerous tumour. Experience in computer aided modelling, coding with Matlab and X-ray imaging analysis is required. The project involved using a micro focus X-ray tube with the state-of-art HEXITEC, an energy-resolved pixelated detector under the imaging setup. Due to the high demand of the detector, major laboratory sessions that involve using the detector are required in the first two months of the project.


X-ray collimation for rapid assessment of surgically removed cancerous tissue
Supervisor: JC Khong and Robert Moss

Breast cancer is the most common cancer in the UK, according to Cancer Research UK, and it accounted for 15% of all new cancer cases in 2015. Surgical removal of the cancerous tissue from the patient is among the highest primary treatment accounting for 81% of patients. A histopathological analysis will be carried out on the tumour tissue to ensure that all of the malignancy has been removed, unfortunately a second surgical procedure will be required if cancerous tissue is found at the margins (edges) of the resected tissue. To avoid secondary surgery, a rapid,
automated analysis on of the excised tissue in the treatment room could be a solution. However, this has to account for the extension of the surgery time, which will increase the risks to the patient. Small Angle X-ray Scattering and Angular and Energy Dispersive X-ray diffraction has been reported to be able to distinguish the differences between normal and cancerous tissue. We are exploring a novel collimator (specially shaped X-ray shielding) that can be used together with an energy-resolved pixelated detector to allow multi- locations to be scanned in parallel. The student is require to design the tissue equivalent sample and compare the obtained result with the energy dispersive imaging technique in another project. Experience in computer aided modelling and coding with Matlab is required. The project involved using a micro focus X-ray tube, a prototype collimator and the state-of-art HEXITEC, an energy- resolved pixelated detector under the diffraction setups. Due to the high demand of the detector, major laboratory sessions that involve using the detector are required in the first two months of the project.


Surgical robotics within the O-arm

Supervisor: Eddie Edwards and Dan Stoyanov

Robotics is revolutionising the way surgery is performed, providing excellent 3D vision and an intuitive interface for surgeons to perform accurate and effective procedures with greater dexterity. At WEISS we have a unique facility in that we can perform lab-based experiments using the O-arm, which provides live fluoroscopic images and also 3D cone beam CT. This project will aim to provide hand-eye calibration between the robot kinematics and X-ray imaging enabling new navigation methods.


Hololens navigation for surgery

Supervisor: Eddie Edwards and Dan Stoyanov

The Hololens provides an amazing immersive environment. This project area aims to examine the use of this device to provide live and accurately aligned images to the surgeon during operations. The aim is to align the coordinate systems of the Hololens and that of an external tracker using novel calibration techniques. Accurate and detailed segmentation of preoperative scans is required and subsequent physical alignment should be achieved - both manually and using automated computer vision techniques. Applications in the prostate, liver and colorectal surgery are possible.


AR guidance and surgical navigation

Supervisor: Eddie Edwards and Dan Stoyanov

We are developing a generic platform for augmented reality image-guided surgery. This allows the surgeon to have a three-dimensional view of data extracted from preoperative images and viewed directly on the patient. Current applications include brain surgery, prostatectomy  and liver surgery, but AR has the potential to improve many other forms of surgery. Picking a particular application, this project will examine the feasiblity of AR guidance, starting from the ability to identify relevant anatomy and pathology in the scans, the possibility and mechanism of accurate alignment to the view of the patient and choice of visualisation method. A propotype AR system should result from this work in collaboration with developers at WEISS.


System for In-Shoe Temperature Monitoring of Feet

Supervisor: Prabhav Nadipi Reddy and Eve Hatten

Diabetic foot ulcers are a major complication of diabetes that reduce the quality of life of people significantly. Monitoring foot temperatures of diabetic people is a possible means of predicting ulceration. This project seeks to develop an insole with embedded temperature sensors that would record foot temperatures of people as they walk. The project involves building a sensor system along with appropriate data acquisition to record and store the data.

This project will suit an engineering student who is interested in electronics, programming, product design and medical device development.

 

Wireless Data Acquisition System for Biomedical Signals

Supervisor: Prabhav Nadipi Reddy and Eve Hatten

The aim of the project is to build a wireless data acquisition system based on an Arduino system that could record different biomedical signals. We are hoping to develop a recording system that can be used as an off-the-shelf product for making different kinds of sensor systems.

The project would suit an engineering students interested in building electronics, programming, product design and medical device development.

 

Estimating Time Spent in Near and Far Work in Children

Supervisor: Prabhav Nadipi Reddy and Eve Hatten

The time spent in doing near work (defined as work in which children are looking at objects less than 50 cms from their eyes) vs. far work is a predictor of prevalence and progress of myopia in children. In this project we want to develop a sensor system to estimate the time children spend doing near and far work. The project would involve selecting sensors, developing appropriate signal conditioning for the sensor and testing the sensor system to prove its efficacy.

The project would suit an engineering students interested in building electronics, programming, product design and medical device development.