XClose

Medical Physics and Biomedical Engineering

Home
Menu

MPBE Summer Studentship Opportunities

Summer Studentships offer a stipend for UCL Medical Physics and Biomedical Engineering students to pursue a project developed in collaboration with an academic over a number of weeks in the summer.

Male and female students working with the Da Vinci robot
Each year the Department of Medical Physics and Biomedical Engineering offers a number of summer placement opportunities for undergraduate students. These placements are an excellent way for students to gain project and research experience during the summer months.

This year, we have two types of studentships available; Research Support and also Teaching and Learning support. Applicants are welcome to apply for 2-3 projects from either category. 

These posts are open to all undergraduate students, including those in their final year and iBSc students. Projects will start in early July 2024.

 

Research Support Summer Studentship

These placements are intended to allow undergraduate students to undertake paid work during the summer with academics. The key aims of the studentships are to give students the opportunity to work on a collaborative research project, gain experience developing and writing a research proposal and presenting their research to their peers.

Please note: Working patterns are categorised as: 'On site' ( >80% of working time), 'Hybrid' (20%-80% on site) and 'Remote' ( <20% on site)

Available projects:

Design and development of training models (phantoms) for robotic-assisted bronchoscopy 

Supervisor: Dr Efthymios Maneas (efthymios.maneas@ucl.ac.uk), Mr Jia-En (Danny) Chen (jia-en.chen.22@ucl.ac.uk) and Prof Adrien Desjardins (a.desjardins@ucl.ac.uk

Working pattern: On site

Project Aim: The aim of this research-centred studentship proposal is to design and develop phantoms for clinical training and pre-surgery simulation in robotic-assisted bronchoscopy through the utilisation of 3D printing technologies. 

Project Summary: 

Technical Background: This project is focused on optimising a new technique developed by the applicants at UCL for fabricating clinical training models compatible with bronchoscopy. The use of 3D printing technology in the medical field has grown significantly in recent years, and there is strong interest in training and skills evaluation in respiratory medicine.  

Location and Resources: This project will be situated in Charles Bell House in the WEISS centre, which has extensive facilities for 3D printing and clinical validation with a mock operating room. Additionally, it will draw upon resources with silicone printing at the Royal Free Hospital and at WEISS with a larger 3D silicone printer that will be installed until the end of June. 

Clinical Collaboration: This project will involve close collaboration with Dr Neal Navani and Dr Ricky Thakrar who are Consultants in Respiratory Medicine at UCLH. The applicants have been collaborating very closely with them for over three years. The project will involve three phases. First, experimental tests will be conducted to assess the impact of different 3D printing materials on the performance of these models during clinical training. Second, the results of these tests will be used to fabricate novel clinical training models. Third, the models will be evaluated by Dr Navani and Dr Thakrar and their colleagues at UCLH, which will involve a combination of inexperienced and experienced clinicians.  

Clinical Impact: Ultimately, the research findings will contribute to advancements in medical training and positively impact patient outcomes.  

Development of a high throughput platform for radiobiology studies using proton radiation in a clinical facility 

Supervisor: Dr Reem Ahmad (r.h.ahmad@ucl.ac.uk), Prof Colin Baker (colin.baker4@nhs.net) Head of Proton Physics at UCLH and Prof Andrew Nisbet (andrew.nisbet@ucl.ac.uk

Working pattern: On site

Project Aim: To design, develop and dosimetrically characterise a high-throughput platform that allows tissue culture plates to be irradiated at the UCLH proton centre. 

Project Summary: This project consists of three parts. In the first, the student will initially consider the geometry and dimensions of tissue culture plates used in radiobiological studies and then design and build a suitable holding device. They will receive support from a designated workshop to build the device however, they will be expected to create adequate schematics to enable this. The device will be made from a readily available material commonly used in radiation experiments, such as Perspex. 

In the second part, the student will obtain a CT scan of their device, setup with each type of plate they plan to test. These will then be imported into a treatment planning system where they will devise a plan. Prior to creating their plan, they will conduct experiments to determine the water equivalent thickness of their material. This is critical for proton experiments, as accurate knowledge of this is needed to ensure the correct delivery of the planned dose. These values will be taken into account with the treatment plan, where necessary. Treatment plans have been already created for different setups, therefore the student will receive sufficient support to achieve the goal of creating deliverable plans. These plans will be sent to UCLH in preparation for the experiment. 

The final part will involve the student conducting experiments at the UCLH proton centre to characterise the properties of their device. Full dosimetric measurements will be conducted in a ‘dry’ run of a radiobiological experiment. This will involve delivery of each plan to each setup in triplicate to allow for some uncertainty measurements. The results from these experiments will be analysed and written up into a short report.  

Development of an ex-vivo Organ Perfusion System for Photoacoustic Imaging Research 

Supervisor: Loïc Lerville-Rouyer: (l.lerville-rouyer@ucl.ac.uk), Fatima Mansour : (f.mansour@ucl.ac.uk), Paul Beard (p.beard@ucl.ac.uk), Rehman Ansari (r.abdul@ucl.ac.uk

Working pattern: On site

Project Aim: Contribute to the development of an ex-vivo organ perfusion system used as a platform to maintain organs alive while performing photoacoustic imaging (PAI). The system is tested on animal and human tissue to assess organ perfusion and viability. The student will work on sub-systems development and integration. 

Project Summary: PAI is a novel imaging technology combining ultrasound and pulsed laser light to collect more information on an organ than just an image. The target would be to detect blood oxygenation level (sO2), a marker than can help detect cancerous cells in hypoxic regions. 

To test PAI, we need to recreate in-vivo conditions in the laboratory, using a perfusion system. 

The perfusion system is made of hardware and software components. It counts about 30 different elements including several in-line sensors (pH, temperature, pressure, pO2, spectrometry, flow) to recreate physiological conditions.  

A schematic representing the system as it is currently set-up can be found below. The three sub-systems are interacting with each other on top of which the Photoacoustic Imaging system is placed to image the liver before and during perfusion. 

Currently, we are perfusing animal organs for 30 min – 1h. With the PAI probe, we try to assess organ perfusion at a sub-surface level during the course of the perfusion. In the long term, the objective would be to correlate measurements from the PAI probe to in-line spectrophotometric sO2 measurements to detect and map tissue hypoxia.  

For this project, the aim is to develop the system further to be able to maintain a viable organ for up to three hours while finding a way to easily acquire images via the Photoacoustic system. Furthermore, other in-line sensors could be integrated in the system to help with organ viability assessment and/or oxygenation measurements. This will involve hands-on work with hardware and software development, integration of in-line spectrometric measurements and other sensors and participation to clinical experiments with organ retrieval and perfusion. 

Extracting and analysing glioblastoma cell fate dynamics from timelapse imaging

Supervisor: Jamie Dean (jamie.dean@ucl.ac.uk), Peter Embacher (p.embacher@ucl.ac.uk

Working pattern: Hybrid 

Project Aim: Establish an image analysis pipeline to study long-term inheritance effects in glioblastoma 

Project Summary: Glioblastoma is one of the most aggressive and difficult to treat cancers. One obstacle to developing effective and widely applicable treatments is the high level of heterogeneity, both between patients and within a single tumour. We aim to better understand how this heterogeneity evolves over several generations of cancer cells. Combining mathematical modelling with quantitative live-cell imaging is a powerful approach to discover mechanisms of variability in the proliferation and therapy response of cells. In order to capture this variability in quantitative measures of cellular proliferation and therapy response, it is vital to be able to reliably process a large number of cells over the course of several generations with minimal error rate. The aim of the project is to develop a computational live-cell imaging analysis pipeline that automatically identifies and tracks the cells to determine their dynamics and genealogical relations. This will involve combining tools from machine learning and statistics to extract the relevant features from microscopy images with the high fidelity necessary to capture intergenerational dependencies. Apart from the image analysis this project will likely also entail modelling aspects to identify and exploit relevant cell characteristics.

A background in programming (preferably Python) and statistical modelling would be helpful. Background biological knowledge is not needed. 

Incorporating spatiotemporal patterns of brain development into paediatric brain tumour radiotherapy planning to reduce treatment side effects 

Supervisor: Jamie Dean (jamie.dean@ucl.ac.uk) and Mohammad Amin Lessan (mohammad.lessan.20@ucl.ac.uk

Working pattern: Hybrid

Project Aim: We aim to understand the spatiotemporal variation in radiosensitivity of the developing brain. This understanding will be exploited to design novel personalized radiotherapy planning approaches to spare the (age-dependent) most radiosensitive regions of the brain. 

Project Summary: Paediatric central nervous system tumours are the second most common childhood malignancy. They are frequently treated with radiotherapy and have relatively high survival rates. However, patients are often left with debilitating neurocognitive side effects resulting from radiation-induced brain damage. The developing brain is particularly sensitive to radiation, with different regions expected to be more or less sensitive at different stages of development. This spatiotemporal variation in radiosensitivity is not accounted for in current treatment approaches. However, technological advances in the spatial targeting of radiotherapy (such as the new UCLH Proton Beam Therapy Centre https://www.youtube.com/watch?v=PD0100l65Qc) provide an unexploited opportunity to spare the most radiosensitive regions of the brain to reduce side effects. In this project we aim to understand the spatiotemporal variation in radiosensitivity of the developing brain. This understanding will be exploited to design novel personalized radiotherapy planning approaches to spare the (age-dependent) most radiosensitive regions of the brain. 

We seek to achieve these aims through the application of machine learning to gene expression data. We will use radiation response data of a large panel of cell lines with matched gene expression data to develop a machine learning model to predict radiosensitivity based on gene expression. We will then apply this model to spatiotemporal gene expression data of the developing brain to predict the radiosensitivity of different regions of the brain at different ages. We shall validate these predictions using matched radiotherapy dose distributions and toxicity data of previously treated paediatric brain tumour patients (time permitting). Finally, we will assess the feasibility of designing radiotherapy plans that spare the (age-dependent) most radiosensitive regions of the brain (future work beyond the scope of this project). We intend that this work will contribute to the design of future clinical trials to determine whether neural development-informed radiotherapy planning can reduce the side effects of paediatric brain radiotherapy. 

Integration of optical ultrasound with a surgical robotic platform for lung biopsy guidance

Supervisor: Dr. Erwin Alles (e.alles@ucl.ac.uk), Dr. Semyon Bodian (zccasbo@ucl.ac.uk), Shaoyan Zhang (shaoyan.zhang.20@ucl.ac.uk)

Working pattern: On site

Project Aim: This studentship will be part of a larger “ROpUS” project aimed at integrating miniature optical ultrasound (OpUS) imaging probes with a surgical robotic platform for real-time imaging. Your contribution will be to carry out tests and optimisation of the imaging performance, and work towards the robotic integration and overall system design. 

Project Summary: Lung biopsies are either performed under CT guidance, which offers limited imaging contrast, or under endobronchial ultrasound (EBUS) imaging. In EBUS imaging, miniature ultrasound piezoelectric transducers are introduced into the bronchi to image the lungs from within, at excellent spatial and temporal resolution and high imaging contrast. However, the size of EBUS probes restricts its application to just the largest bronchi, meaning that narrower bronchi in the lower half of the lungs are inaccessible and thus, unexplored.  

To allow for EBUS imaging at greater depths, highly miniature OpUS imaging probes and real-time imaging console have been developed within UCL. An OpUS device generates and receives ultrasound waves to using laser light channelled via optical fibres, resulting in highly miniaturised ultrasound transducers that offer high-resolution ultrasound imaging and immunity to electromagnetic interference. The OpUS ultrasound probes will be integrated with a surgical robot (the Intuitive Ion platform) for accurate delivery. A separate, high-speed linear actuator will control the OpUS probe for real-time imaging to guide lung biopsies.  

The project will consist of three aspects. First, you will be tasked with designing the tools and structures to integrate the OpUS probes with the robotic platform and actuator. The student will also carry out essential tests of the resulting devices. Second, the student will have the opportunity to develop phantoms for imaging performance characterisation. Third, the student will be exploring the methods of optical fibre processing, such as cleaving, housing, and coating for imaging optimisation. 

This work is predominantly experimental in nature, and will involve rapid prototyping techniques (e.g. laser cutting, 3D printing, training provided) to fabricate structures and components for robotic integration as well as, imaging phantoms. In addition, the project will involve; processing and handling of optical fibres (training provided). Previous laboratory experience would be preferred but is not essential. 

Magnetic Resonance Fingerprinting of Magnetic Field Perturbations 

Supervisor: Patrick Fuchs (p.fuchs@ucl.ac.uk)

Working pattern: Hybrid

Project Aim: To apply a dictionary fitting approach from magnetic resonance fingerprinting (MRF) to the magnetic field mapping problem in quantitative MRI susceptibility mapping (QSM). We aim to replace current state-of-the-art non-linear fitting methods with MRF dictionary matching which should be more robust to noise. 

Project Summary: QSM is a maturing MRI technique with emerging clinical applications because tissue magnetic susceptibility is related to its composition including the content of iron, myelin, and calcifications. The first step in conventional QSM is to calculate maps of magnetic field perturbations (ΔB) from the complex MRI signal. This is usually done by nonlinear fitting of the complex signal over multiple echo times but, here, we propose to exploit MRF which allows highly efficient tissue parameter mapping by dictionary matching. This involves simulating the MRI complex signal evolution for a large range of tissue parameters, creating a large dictionary of simulated signals, and matching this dictionary to the measured signal to yield parameter estimates. 

For QSM, you will develop a dictionary of phase evolutions over echo times, related to the underlying tissue transverse relaxation rate (R∗2), proton density (ρ), magnetic field perturbation (ΔB), and phase offset (ϕ0). You will then use this dictionary to “fit”, by MRF dictionary matching, the measured complex signal over multiple echoes in each voxel to obtain the underlying parameter maps. 

This dictionary can then be compressed or simplified depending on the practical range of the parameters as well as their relationships. These relationships can also be used to make the matching approach more robust to noise. For example, large observed magnetic field perturbations may suggest high transverse relaxation rates both resulting from large magnetic susceptibilities. 

Finally, spatial relationships may be captured in the dictionary if matching is done within volumes or patches including many voxels rather than within each voxel separately. In this case, generating a dictionary of volumes would allow it to capture the smoothness of the phase offset map, and to reflect the nonlocal spatial relationship between microscopic susceptibility sources (both paramagnetic and diamagnetic), magnetic field perturbations, and tissue relaxation rates. 

Modelling MR-guided focused ultrasound treatments for essential tremor 

Supervisor: Prof Bradley Treeby (b.treeby@ucl.ac.uk)

Working pattern: Hybrid

Project Aim: The aim of this project is to develop a transducer model for the Insightec Exablate Neuro and integrate this with the k-Plan treatment planning software to allow treatment planning simulations for MR-guided focused ultrasound treatments for essential tremor.

Project Summary: Magnetic Resonance-guided Focused Ultrasound Surgery (MRgFUS) represents a cutting-edge therapeutic approach for treating essential tremor, a common movement disorder characterized by involuntary and rhythmic shaking that significantly impacts the quality of life. MRgFUS leverages focused ultrasound waves to create precise lesions in the brain tissue, specifically targeting areas responsible for the tremor. The integration of Magnetic Resonance Imaging (MRI) ensures real-time imaging guidance and temperature monitoring. This non-invasive technique offers a promising alternative to traditional surgical interventions, reducing the risk of complications and promoting faster recovery. 

However, the effectiveness and safety of MRgFUS depend critically on precise treatment planning and execution. Current methodologies, while effective, can benefit from enhancements in simulation capabilities to predict the acoustic and thermal effects of ultrasound waves within the complex anatomy of the human brain. Model-based treatment planning can offer significant improvements by enabling the simulation of various treatment scenarios, optimizing transducer positioning, and predicting tissue responses before the actual procedure. This approach can potentially increase the accuracy of targeting, minimize damage to surrounding tissues, and improve overall treatment outcomes. 

The primary aim of this project is to integrate a sophisticated transducer model of the Insightec Exablate Neuro into the k-Plan treatment planning software package. The model parameters will be defined based on information provided by the manufacturer. A key step will be computing the coordinate system transformations needed to position the transducer relative to the planning images. Depending on the time available, the next step will be to compare simulation results with outcomes from previously completed treatments (temperature maps measured using MR-thermometry during patient treatments). Ultimately, this project aims to enhance the precision, safety, and efficacy of MRgFUS treatments, paving the way for more personalized and optimized therapeutic strategies for patients suffering from essential tremor. 

PitSAM: Low-Rank Adaptation of SAM for Pituitary Adenoma Segmentation from MRI  

Supervisor: Dr Mobarakol Islam (mobarakol.islam@ucl.ac.uk) and Prof Matt Clarkson (m.clarkson@ucl.ac.uk

Working pattern: Hybrid

Project Aim: We aim to design an efficient adaptation technique of segment anything model (SAM) using text prompt for the pituitary tumour segmentation from MRI for diagnosis and augmented reality-based localization and tracking during endonasal surgery.  

Project Summary: The automatic segmentation of the pituitary tumour from MRI is important for surgical planning and augmented reality-based localization and tracking during endonasal surgery. However, manual segmentation of the pituitary tumour is expensive, time-consuming, and inclined to the inter-rater error. In addition, the unavailability of the datasets and labels are the major constraints to developing deep learning-based automatic segmentation models to support the treatment. The foundation model, segment anything model (SAM) [1], is showing state-of-the-art zero-shot performance in the vision tasks. However, recent works [2,3,4] investigate that SAM requires adaptation with domain-specific tasks, especially in medical image segmentation. In this work, we aim to design an efficient adaptation technique of segment anything model (SAM) using text prompt for the pituitary tumour segmentation from MRI. 

Datasets: (i) our in-house pituitary tumor segmentation dataset; (ii) publicly available dataset of brain tumor segmentation[12].  

Requirements: Python, Pytorch, Deep Learning  

References: 

[1] Kirillov, Alexander, et al. "Segment anything." arXiv preprint arXiv:2304.02643 (2023).  

[2] Zhang, Kaidong, and Dong Liu. "Customized segment anything model for medical image segmentation." arXiv preprint arXiv:2304.13785 (2023). 
[3] Wu, Junde, et al. "Medical sam adapter: Adapting segment anything model for medical image segmentation." arXiv preprint arXiv:2304.12620 (2023).  

[4] Hu, Xinrong, Xiaowei Xu, and Yiyu Shi. "How to Efficiently Adapt Large Segmentation Model (SAM) to Medical Images." arXiv preprint arXiv:2306.13731 (2023). 

[5] https://www.med.upenn.edu/cbica/brats/  

Prostate cancer lesion prediction (with whole prostate masking) from bi-parametric MRI using deep learning 

Supervisor: Dr. Laxmi Muralidharan (l.muralidharan.11@ucl.ac.uk)

Working pattern: Hybrid

Project Aim: The project goal is to increase the contouring performance for cancerous lesions in the prostate by training a nnUNet segmentation network along with an EfficientNet-b5 classification network. Multi-site bi-parametric data set along with whole prostate masks, predicted by deep learning, and clinical variables will be utilised for the training to investigate the generalisability of training on challenge data to an unmatched clinical acquisition. 

Project Summary: Prostate MRI is extremely useful in the early diagnosis of clinically significant prostate cancer (csPCa) and helps to reduce unnecessary interventions or treatments in patients with benign lesions or indolent prostate. However, it is time-consuming and difficult for radiologists to detect and contour csPCa on multi-parametric MRI images. Deep learning has shown remarkable performance in many image processing tasks but requires training with large databases of annotated images. 

Clinical bi-parametric prostate MRI (bpMRI) data includes axial T2-weighted images, diffusion-weighted images (DWI) with a high b-value and apparent diffusion coefficient (ADC) maps calculated from DWI. PI-CAI (Prostate Imaging: Cancer AI) is a 2022 grand challenge encompassing over 10,000 carefully annotated prostate MRI scans. This large-scale publicly available challenge dataset allows researchers to design, train and test publicly available deep learning models. Additionally, 13 clinical datasets with csPCa  lesions contoured by a radiologist from a UCLH prostate cancer study (HistoMRI) are available.  

In this project, you will use the PICAI data along with 8 clinical datasets from the histo-MRI study to train deep learning models to identify csPCa on bi-parametric prostate MRI. You will then test and evaluate the trained networks on 5 clinical data sets from the Histo-MRI study. This will help investigate the usability of large scale challenge dataset trained models to small clinical trials. 

Although deep learning models based on segmentation networks nnUNet and nnDetection performed very well for lesion segmentation in the PI-CAI challenge, a model that combined a 2D segmentation network (ITUNet) with a classification network achieved highest scoring on the validation set in the open development phase. Therefore, here we propose to use nnUNet for 3D segmentation together with a classification network (EfficientNet-b5) to improve csPCa lesion prediction performance.  

Testing novel optical sensors for high intensity focused ultrasound systems 

Supervisor: Elly Martin (elly.martin@ucl.ac.uk) and Ben Cox (b.cox@ucl.ac.uk)

Working pattern: On site

Project Aim: The aim of this project is to experimentally test novel holographic sensors that have been shown to respond to the acoustic fields from high intensity focused ultrasound devices in order to (1) better understand the mechanism by which they work, and (2) calibrate them so they can be used to characterise HIFU fields. 

Project Summary: High Intensity Focused Ultrasound (HIFU) also called Focused Ultrasound Surgery (FUS) is a therapeutic use of ultrasound for treating a range of conditions, including uterine fibroids, essential tremor, and tumours. HIFU is also used in physiotherapy to warm tissues, and there are even cosmetic uses of HIFU, eg. for skin tightening. HIFU works by thermally coagulating the tissue through the absorption of the ultrasound energy. If used incorrectly, with the wrong acoustic intensity for example, it could cause unwanted irreversible tissue damage. It is therefore critical that the acoustic output of HIFU devices is checked prior to its application.  

This project will be in collaboration with Dervil Cody from TU Dublin.

She and her team have produced holographic optical devices that are affected by high intensity ultrasound fields and could potentially be used to measure the output of HIFU devices and assess their safety, but the mechanism by which the ultrasound affects the devices is not well understood. 

In this project, to explore what the mechanism might be, we propose to insonify the novel sensors with focused ultrasound fields from a range of transducers in order to evaluate the sensor response as a function of frequency, acoustic intensity, and insonication time. The results from these initial measurements will indicate the most likely mechanism (cavitation, heating, radiation pressure,…) and a second stage of experiments will be designed and conducted to confirm this. Based on this improved understanding, a series of measurements will be undertaken to calibrate the response of the sensors. 

The research will be carried out in the laboratory of the Biomedical Ultrasound Group (3rd floor Malet Place Engineering Building). 

Understanding sequencing results from patient data with protein misfolding calculations 

Supervisor: Ben Hall (b.hall@ucl.ac.uk)

Working pattern: Hybrid

Project Aim: Develop production web app for analysing mutation data through misfolding calculations.

Project Summary: 

Protein misfolding is a common mechanism by which variants to a gene can lead to disease. In several genetic diseases, misfolding has been used successfully as a tool to predict clinical outcomes. However, performing these calculations requires expertise and access to high performance computing facilities, limiting their wider adoption. 

Our team has developed a computational workflow for precalculating the impact of misfolding, performed a systematic study of how to integrate estimates taken from different experimental sources, and developed prototype web apps and APIs for interacting with databases of precalculated results. Your job will be to integrate these into a single web app that allows users to input gene variant data and retrieve analyses and interpretations of the mutations based on misfolding calculations. 

Self-supervised deep learning to improve quantification of blood-brain barrier permeability from MRI data 

Supervisor: Dr Elizabeth Powell (e-powell@ucl.ac.uk); Prof Geoff Parker (geoff.parker@ucl.ac.uk); Snigdha Sen (snigdha.sen.20@ucl.ac.uk

Working pattern: TBC

Project Aim: To use a self-supervised deep neural network to improve the quantification of BBB permeability. 

Project Summary: The blood-brain barrier (BBB) provides a protective layer between blood vessels and brain tissues, preventing any blood borne toxins from entering the brain. Damage to the BBB affects this seal around the blood vessels so that they become “leaky,” meaning that toxins are able to cross into the brain. The rate of water exchange across the BBB is emerging as one of the most promising MRI biomarkers for quantifying its permeability, but current acquisition protocols suffer from low image quality. This means that standard parameter estimation techniques using non-linear least squares approaches using this low-quality data suffer from noise and biases. Deep learning, in particular self-supervised learning, has been shown to improve parameter estimation by reducing these biases, whilst also providing significant gains in speed. 

In this project, the student will implement a self-supervised deep neural network to improve the quantification of BBB permeability. Improvements in parameter estimation will first be evaluated using simulations; the technique will then be applied to a pilot data set of healthy volunteers. 

Teaching and Learning Support Summer Studentship

These placements are intended to allow undergraduate students to undertake paid work this summer with academics to develop teaching and learning materials, content or activities for use in MPBE courses.

Please note: Working patterns are categorised as: 'On site' ( >80% of working time), 'Hybrid' (20%-80% on site) and 'Remote' ( <20% on site)

Available projects:

Developing an app to help students with practical electronics. 

Supervisor: Prabhav Reddy (p.reddy@ucl.ac.uk) and Eve Hatten (e.hatten@ucl.ac.uk)  

Working pattern: TBC

Project Aim: To create resources to help students understand practical electronics equipment and look at building an app to help students detect issues when building circuits on breadboards.  

Project Summary: In this project a student will have two main focuses:  

 

  1. Creating short videos to teach students how to set up and adjust common lab equipment.  
  2. Developing an app that will help a student to spot errors in their breadboard electronic circuits. 
    The idea here is to use a photograph of the breadboard circuit and use image processing to convert the image to a netlist which can be compared with a circuit diagram. 

 

Ideally both of these will be combined into one app that we can offer to undergraduate Biomedical Engineering students to support their electronics learning.  

The ideal student for this project is a Biomedical Engineering student with a good understanding of electronics, with an interest in machine learning and app development.  

 

How well do our healthcare related programmes prepare students for employment? 

Supervisor: Pilar Garcia Souto (p.garciasouto@ucl.ac.uk) and Arsam Nasrollahy Shiraz (a.shiraz@ucl.ac.uk)

Working pattern: TBC

Project Aim: The primary aims are to identify key employers within healthcare engineering (HE) industry, the technical skills they sought after, and courses at UCL that provide students with a path for acquiring such HE specialised skills.

IHE EdDG previously identified and surveyed HE related modules and this summer studentship work will allow us to assess current UCL strengths and areas of improvement.  

Project Summary: One of the aims of IHE EdDG is to provide a link between various providers of HE specialised content and skills across UCL to facilitate the provision of HE specialised content. We currently hold a record of HE related modules at UCL which was collected through a project within IHE EdDG. We would like to utilise the information we currently have to investigate in which HE areas UCL is strong and in which areas UCL is weak considering quantitative and qualitative data on industry requirements with respect to HE specialised skills. 

This will inform/help us to identify the potential for interdepartmental collaboration on HE education or need to introduce new modules or courses. 

Within this project the student will: 

(1)  identify potential graduate destinations within HE industry. This will involve a thorough search of current and past vacancies. The data will be organised and stored for future analysis in terms of company’s activities and person’s specifications.  

(2) For high potential employers and other companies with a UCL alumnus as an employee we aim to gather further information by contacting the respective UCL alumni. The summer studentship will ask UCL Alumni working in a HE related field to complete a survey, and will also carry out a series of interviews. The survey will be developed in collaboration with the IHE EdDG members, but will hope to investigate how the different HE related programmes have prepared UCL Alumni for employment, strengths and areas of improvement. 

(3) Prepare documentation summarizing findings. This includes a document ready to share with students seeking information about potential industry destinations, and also a report that will be shared with key staff in departments. 

Teaching resources for MRI-guided robotic devices

Supervisor: Ziyan Guo (ziyan.guo@ucl.ac.uk)

Working pattern: Hybrid

Project Aim: To develop teaching resources for MRI-guided robotic devices, including examples in course notes and designing a platform for material testing. 

Project Summary: Magnetic resonance imaging (MRI) is a non-invasive imaging modality that can offer high-contrast three-dimensional images of soft tissue without harmful radiation. It can also provide functional information, including flow measurement, and molecular information, particularly in central nervous system assessment. Its versatility has led to its use in a wide range of research and clinical scenarios, such as neuroscience, diagnosis and treatment guidance for prostate biopsy, brain surgery and cardiovascular interventions. In this context, MRI-guided devices and robots are drawing more and more attention from both academia and industry.  

MPHY0052 MRI-guided Devices, a new module that will be taken by MSc students, will provide a comprehensive review of medical devices and robotics for MRI-guided procedures, and allied technologies, for instance, new materials and sensing techniques.  

In this summer internship, the student will design a remotely controlled platform for material testing, which should be compatible with the existing MRI machines. if successful, the student will have the opportunity to see their designs implemented.  

This project will involve computer-aided design (CAD) modelling, 3D printing, CNC manufacturing and position control (Matlab).  

Use of Large Language Models for Teaching Electronics in an University Setting 

Supervisor: Prabhav Reddy (p.reddy@ucl.ac.uk) and Eve Hatten (e.hatten@ucl.ac.uk)

Working pattern: Remote

Project Aim: The aim of the project is to fine-tune a large language model such that it can answer simple questions about theory and practice of electronics as taught in the department. The aim is to create an app based chatbot for the students to engage with. 

Project Summary: Large Language Models are quite good at understanding and answering unstructured questions. Though these models (eg. chatGPT) are good at answering general questions, they are very bad at answering questions on electronics. These models can be fine-tuned so that they respond appropriately. The fine-tuning would enable to models to answer questions on electronics and disregard irrelevant questions. 

The project would involve selecting an appropriate open-source large language model that can be fine-tuned, creating a database of training data to use for fine-tuning of the model, training the model on the data, and testing the model. Further, we want to  think about ways in which such LLMs might be useful in university teaching including the ethics and ways in which we can prevent their misuse. 

The project would require the student to have experience in python programming. Prior experience in machine learning is desirable though not required. 

Understanding the challenges that students face in university education

Supervisor: Adam Gibson (adam.gibson@ucl.ac.uk), Rebecca Yerworth (r.yerworth@ucl.ac.uk), Eve Hatten (e.hatten@ucl.ac.uk) and Jenny Griffiths (Arena, j.griffiths@ucl.ac.uk)

Working pattern: Hybrid

Project Aim: 1. To carry out interviews with undergraduate students to understand what additional support they feel they 
would have benefitted from as they started at UCL
2. To co-design a week long programme aimed at preparing students for university education

Project Summary: We know that students report challenges in transitioning to university education, which is reflected in 
lecturers’ comments about the perceived low ability of students. The redesign of the biomedical 
engineering programme allows for some flexibility in introducing an additional induction period aimed at 
transitioning students into university.


We aim to employ a current undergraduate student and train them to deliver a questionnaire and 
interviews to biomedical engineering students. The questionnaire and interviews will aim to understand 
what students feel about the transition to university. What do they feel were the main challenges? What 
do they wish they had been told when they started university?


We have the opportunity to modify the induction process in the redesigned biomedical engineering 
programme, to offer additional support to first year students. We could use this to explain and reinforce 
expectations of commitment and behaviour, but for this to be effective, we need to learn from students. A 
peer-led questionnaire and interview process should highlight issues that students face, which we can 
incorporate into introductory sessions.


We will also work with the student to co-design activities that best meet the students’ needs, with the aim of 
preparing students for lectures, coursework, assessment and projects.


Eligibility

Open to all undergraduate students in Medical Physics and Biomedical Engineering, including iBSc and those in their final year (due to graduate this summer).

Renumeration

Each studentship is accompanied by a stipend of £3,500 for an 8-week project (pro-rated for shorter projects), typically commencing at the start of July. Training will be provided appropriate to the requirements of the project. In addition, up to £200 may be requested for materials and consumable items.

Submission Instructions

Submit your application

  1. You are required to complete the above application form.
  2. You are required to submit a cover letter (no longer than one page) and specify at the top which are your 2 or 3 projects in order of preference. In your cover letter, you should address how you meet the required attributes, your prior experience in research and/or teaching and why you are interested in the chosen project/s.
  3. You must also submit a one-page CV.

DEADLINE: Sunday 14th April 2024, 23:59

If you have any questions, please email medphys.teaching@ucl.ac.uk