PhD Projects for GTA Applicants - September 2019

For information about the GTA posts and how to apply, click here.

If you apply for one of our Graduate Teaching Assistantships for September 2019 start, you need to name potential PhD supervisors and submit a research proposal as part of your application:

  • Any member of academic staff (Lecturer, Senior Lecturer, Reader, Associate Professor, Professor) in the Department can be your supervisor.
  • Any research topic related to the departmental research themes can be your PhD topic.

Some examples of exciting PhD projects for which supervisors are currently looking for students are:

Information about these PhD projects is provided below. Further projects may be added in the coming days. If you apply for a GTA post, you can apply for one of these projects, but you can also propose any other member of academic staff as supervisor and propose any research topic related to the departmental research themes as your PhD topic.

A Predictive Analytics Model for University Admission

Supervisor: Dr Yi Hong

The aim of this project is to develop an innovative approach and predictive analytics model to forecast and increase student recruitment at the University of Leicester. The proposed model will unlock the potential of student applicants/enrolment data gathered from internal and external sources (UCAS open days, outreach activities, social networks) using visual analytics and various statistical machine-learning techniques.  This research will be centred around recent development in Artificial Intelligence, including:

  • Descriptive statistics:  distribution analysis and the degree of variable dependence in relation to the enrolment probability;
  • Relationship modelling: This includes identification of suitable statistical techniques for relationship modelling, discovery and predictive analytics;
  • Verification of results: the prediction and classification accuracy will be evaluated by examining historical traces of student behaviours and their corresponding outcome;
  • Development of algorithm:  a machine learning algorithm and a proof-of-concept visualisation platform will be implemented to demonstrate our methodology.

The outcomes of the project will enable the University Marketing Team to

  • Develop more effective marketing campaigns, reach specific applicants based on the aggregated probability of their demographic groups enrolling at the University of Leicester;
  • Understand how prospective students make choices, the predictive aspects of the proposed system will allow the University to obtain a more accurate picture of expected student numbers.

Please e-mail Dr Yi Hong ( for further information about this project.

Automated Accessibility Testing of Mobile Apps

Supervisors: Dr José Miguel Rojas and Prof. Mohammad Mousavi

An estimated 15% of the world population live with some form of disability and must face multiple barriers in their day-to-day life. In the UK, two million people suffer from sight problems (e.g., blindness, low vision, or colour blindness) and the number is predicted to rise to over 2,250,000 by 2020, according to the Royal National Institute of Blind People. In a society increasingly populated by computer systems, enabling these users to access computer technology effectively is a major concern. According to the Ofcom’s 2015 Communications Market Report, smartphones are the most popular device to access the Internet in the UK, with a third (33%) of British internet users preferring their smartphone over their laptops for going online. This scenario sets an important challenge to the software industry: We must produce mobile apps that not only satisfy basic functional requirements, but also support the population of users with accessibility requirements.

Even when software vendors and developers are aware of accessibility needs, there is an evident lack of tool support to develop accessible apps or assess existing apps’ accessibility. While some accessibility properties can be checked statically, modern development practice indicates that mobile user interfaces are often created dynamically and therefore are not amenable to static checking. And while basic accessibility checking frameworks exist to analyse accessibility properties or mitigate accessibility limitations at runtime, they either require substantial additional effort from developers or are simply not suitable for current mobile development processes. The main aim of this project is to overcome these issues through the use of automated test generation to analyse, assess, and improve the accessibility of mobile apps.

The research objectives for this project are:

  • Develop a fully automated test generation approach to check the accessibility of mobile Android apps.
  • Evaluate the usefulness of the approach by assessing the accessibility of existing apps and interacting with their developers.
  • Evaluate the severity and relevance of the accessibility issues identified with the automated approach, as well as issues missed, using observational studies and interviews with real users with accessibility needs.

Additional reading:

  • Marcelo Medeiros Eler, José Miguel Rojas, Yan Ge, Gordon Fraser. “Automated Accessibility Testing of Mobile Apps,” IEEE Intl. Conference on Software Testing, Verification and Validation (ICST), IEEE, 2018.
  • Alberto Oliveira, Marcelo Eler. “Strategies and Challenges on the Accessibility and Interoperability of e-Government Web Portals: A Case Study on Brazilian Federal Universities,” IEEE Computer Software and Applications Conference (COMPSAC), IEEE, 2017.
  • Camila Silva, Marcelo Medeiros Eler, Gordon Fraser. “A survey on the tool support for the automatic evaluation of mobile accessibility”, Intl. Conference on Software Development and Technologies for Enhancing Accessibility and Fighting Info-exclusion (DSAI), ACM, 2018.
  • Ermira Daka, José Miguel Rojas, and Gordon Fraser. “Generating Unit Tests with Descriptive Names Or: Would You Name Your Children Thing1 and Thing2?” In ACM Int. Symposium on Software Testing and Analysis (ISSTA), ACM, 2017.
  • José Miguel Rojas, Mattia Vivanti, Andrea Arcuri, and Gordon Fraser. “A detailed investigation of the effectiveness of whole test suite generation.” Empirical Software Engineering (EMSE), Springer, 2016.
  • H. Beohar and M.R. Mousavi. “Input-Output Conformance Testing for Software Product Lines.” Journal of Logic and Algebraic Methods in Programming, 85(6). Elsevier, 2016.

Automated Delineation of Stroke Lesions in Medical Images

Supervisors: Dr. Huiyu Zhou, Prof. Thomas Erlebach and Prof. Leong Ng

Project informationClick here (PDF)

Automated Testing in Model Transformation

Supervisors: Dr Artur Boronat and Dr José Miguel Rojas

Software models enable the abstraction of relevant features of a software system. For example, the domain model underlying a microservice helps in determining how to design a scalable database, the model of a system at run time can facilitate monitoring and optimization of non-functional requirements, the model of the software architecture of a system plays a key role in managing release deployment. When using appropriate modelling notations and tools [1], such models become software artefacts that can be processed and analysed automatically using model transformations.

Modern transformation languages, like YAMTL [2,3,4], assist in developing model manipulation tasks involving very large models by helping the developer choose from advanced features for model transformation and from existing functionality in mainstream object-oriented programming languages. As any other software artefacts, model transformations can contain defects, and the likelihood of bugs increases as the models to be processed are used to derive more fine-grained, low-level artefacts, e.g., code. Testing these model transformations is therefore important to validate the correctness and reliability of the resulting software system. Automated testing involves the development and use of techniques and tools [5] that can alleviate the often large costs (e.g., manual inspection and bug detection/fixing) associated with the manual validation of models' behaviour. For instance, search-based test generation techniques can be adopted to automatically generate executable tests to check the conformance between a model and a corresponding code component [6].

In this project, we aim at exploring techniques successfully used in testing of object-oriented programs to verify model manipulation tasks involving model transformation features.

[1] Douglas C. Schmidt. Guest Editor's Introduction: Model-Driven Engineering. IEEE Computer 39(2): 25-31 (2006)
[2] Artur Boronat. YAMTL home page:
[3] Artur Boronat. Offline Delta-Driven Model Transformation with Dependency Injection. FASE 2019: 134-150
[4] Artur Boronat. Expressive and Efficient Model Transformation with an Internal DSL of Xtend. MoDELS 2018: 78-88
[5] Muhammad Shafique, Yvan Labiche. A Systematic Review of Model Based Testing Tool Support. Carleton University, Technical Report, May 2010.
[6] José Miguel Rojas, Mattia Vivanti, Andrea Arcuri, Gordon Fraser. A detailed investigation of the effectiveness of whole test suite generation.  Empirical Software Engineering 22(2): 852-893 (2017).

Differences in Specifications

Supervisor: Dr Jan Oliver Ringert

Project information: Click here (Web Link)

Efficient Co-Evolution of Software Architectures and Artificial Intelligence Components

Supervisor: Dr Jan Oliver Ringert

Project information: Click here (Web Link)

Explainable AI and Synthesis

Supervisor: Dr Jan Oliver Ringert

Project information: Click here (Web Link)

Exploring Design Opportunities for Context-Adaptive Collaborative Tools to Support and Enhance Drone-assisted Search and Rescue (SAR) Operations

Supervisors: Dr Nervo Verdezoto and Dr Michael Hoffmann

Project informationClick here (PDF)

Exploring the Design Space for Enhancing Complementary Feeding Practices in Peru

SupervisorDr Nervo Verdezoto

Project informationClick here (PDF)

Fully optimized deep neural network for computer-aided cancer detection

Supervisors: Prof. Yu-Dong (Eugene) Zhang, Prof. Thomas Erlebach, Dr Reza Zare

Computer aided cancer detection and diagnosis (CADx) has made significant strides in the past decade, with the result that many successful CADx systems have been developed. However, the accuracy of these systems still requires significant improvement, so that they can meet the needs of real-world diagnostic tasks.

Detection of the earliest stages of cancer is fraught with high false positive rates necessitating additional testing, leading to increased patient anxiety, potential over treatment and unnecessary additional costs. The false positivity of cancer detection, coupled with a 2-week waiting time, plus the potential for over treatment and patient anxiety present an ideal opportunity for smarter, faster, cheaper diagnostics. Reducing the time patients wait for a diagnosis from weeks to mere hours and accelerating the introduction of treatment options could have huge implications on cancer care, including reduced patient anxiety and unnecessary treatment, culminating in huge cost savings. Improvements in precision and efficiency mean fewer human errors, leading to a decrease in the length and frequency of follow up visits. Doctors will also be able to get information from data for patients who are at risk of certain diseases to prevent hospital re-admissions.

To improve the accuracy and speed of pathological diagnosis, while potentiating reduced costs, this project will utilise a new deep learning algorithm, particularly an enhanced DenseNet , to more accurately distinguish utility edges in freely-available online images of benign and malignant cancers (coupled with verified pathologic information), ultimately providing enhanced classification of the earliest stages of cancer.

Please e-mail Yu-Dong Zhang (Eugene) for further information about this project. The contact email is:

Additional Information: Click here (PDF)

Intelligent Automated Interpretation and Reporting of Medical Images

Main supervisor: Dr Reza Zare
Co-supervisor: Prof Yu-Dong Zhang (Department of Informatics)
Co-supervisor: Dr Emma Chung (Department of Cardiovascular Sciences)

Project information: Click here (PDF)

Safety Analysis for Autnomous Vehicles

Supervisors: Prof. Mohammad Mousavi and Dr. José Miguel Rojas

Establishing trust in autonomous vehicles is a major component of their widespread public adoption. Rigorous and explainable safety analysis of autonomous vehicle functions play a major role in establishing trust. There are existing standards for establishing safety in automotive systems, of which the ISO 26262 standard is the most prominent one. Safety case analysis in these standards involves defining a safety item and analysing and providing a safety case for the item by analysing the hazards in typical scenarios of use and foreseeable misuse. To analyse the hazards rigorously, different safety integrity levels (ASIL) are attached to them, and different analysis techniques are prescribed for different ASIL. At high ASIL, formal verification and model-based testing are recommended as appropriate techniques for the analysis.

Hitherto, much of the safety case analysis process has been manual, involving tedious scrutiny of possible scenarios and turning them into appropriate models for further analysis. For autonomous vehicles, however, such a manual process becomes extremely laborious and error-prone and mechanisation support is inevitable. Also, in the presence of adaptive and AI-enabled systems, adapting safety cases and their analysis should inevitably be mechanised or otherwise will be infeasible. This project aims at providing mechanised support for safety-case analysis of automated and autonomous functions.

To this end, we will build upon our past experience with automated test-case and scenario generation to turn structured English safety case and item descriptions into rigorous models from which use and misuse scenarios are generated automatically.

Please do not hesitate to contact Mohammad Mousavi ( for further information.

Additional information: Click here (PDF)

The use of deep learning in cell line authorisation for modern drug discovery

Supervisors: Dr Huiyu Zhou, Dr Yinhai Wang, Prof Thomas Erlebach

Project information: Click here (PDF)

Vocal Emotions Analysis of Conversations for Supporting Emotional Wellbeing

Supervisors: Prof. Effie Law and Dr Huiyu Zhou

Project information: Click here (PDF)

Share this page: