Theses

For Students: If you want to write research under my supervision, please send a cover letter, CV, and a one-page research proposal.

Objectives:

  • Learn more about a specific recent scientific topic and a real-world challenging problem.
  • Learn more about academic research and academic writing.
  • Learn more about scientific problem-solving.
  • Learn more about data visualization, statistics, and related tools e.g. Latex.
  • Learn more about presentations and academic talks.
  • If you do well, we can write a publication and:
        • you may present your work at a scientific conference.
        • you may get a full-time PhD paid job in our group (or somewhere else).

Completed Theses:

Master:

  • 2022, Li Ying Yin Simon, 3D Mesh Alignments Using Multi-resolution Non-rigid Image Registration Techniques.

Bachelor:

  • 2016, Robert Julien Claude Kohnen, Cochlea Spiral Shape Detection.

Open Theses (Bachelor/Master):

The main research areas are AI, Data Privacy, Medical Imaging, Cellular Automata, and Biomechanic simulation. You will be given a research problem and tasks based on your level i.e. Bachelor or Master.

You are welcome to suggest your own topic, I may accept if it is related to my research interest.

1. Data Privacy: Data privacy is crucial for protecting sensitive personal information and maintaining trust between users and others e.g. service providers and researchers. Currently a student can work on one of these sub-topics:

  • 1.1 Privacy by Design for AI Systems:
        • Introduction: In this topic, you will help develop a Privacy-by-Design Framework for Enduser-Oriented AI-enabled Systems.
        • Objectives:
                • Learn and study the basics of data privacy and privacy by design.
                • Identification of formal specification of the data protection requirements.
                • Mapping the different levels of criticality in a conceptual model.
                • Compare to available related work.
        • Resources:
                • EU GDPR: https://gdpr-info.eu/issues/privacy-by-design
                • EU AI Act https://artificialintelligenceact.eu
                • ML Attack Models: Adversarial Attacks and Data Poisoning
                  Attacks https://arxiv.org/pdf/2112.02797
                • Thibaud Antignac, Riccardo Scandariato, and Gerardo Schneider. A Privacy-Aware Conceptual Model for Handling Personal Data. In 7th International Symposium on Leveraging Applications of Formal Methods, Verification and Validation - ISoLA'16 (1); Track: Privacy and Security Issues in Information Systems, volume 9952 of LNCS, pages 942-957. Springer, 10-14 October 2016. (pdf)
  • 1.2 Synthetic Medical Data Generation:
        • Introduction: Around the globe, researchers are working on projects aimed at enhancing the availability of data for research. While data sharing continues to be a critical aspect, the development of suitable methods for medical data is challenging, due to the specific sensitivity and uniqueness of medical data. This leads to a dilemma, as there is a lack of both methods and the necessary data to create appropriate approaches in the first place. This topic tries to bridge this gap, by providing synthetic datasets that can form the foundation for such developments.
        • Objectives:
                • Learn and study the basics of medical data and synthetic data generation.
                • Extending our SynthMD tool to work with different data types e.g. nested EHR and genetic data.
                • Investigate the use of different approaches for synthetic medical data generation e.g. using AI.
                • Compare to available related work.
        • Resources:
                • Al-Dhamari et al, (2024). Synthetic datasets for open software development in rare disease research. Orphanet J Rare Dis 19, 265 (2024). https://doi.org/10.1186/s13023-024-03254-2
                • https://github.com/iaBIH/synth-md
                • Walonoski et al. (2018). Synthea: An approach, method, and software mechanism for generating synthetic patients and the synthetic electronic health care record, Journal of the American Medical Informatics Association, Volume 25, Issue 3, March 2018, Pages 230–238, https://doi.org/10.1093/jamia/ocx079
                • Li J, Cairns BJ, Li J, Zhu T. (2023). Generating synthetic mixed-type longitudinal electronic health records for artificial intelligent applications. NPJ Digit Med. 2023 May 27;6(1):98. doi:10.1038/s41746-023-00834-7. PMID: 37244963; PMCID: PMC10224668
                • Mosquera, L., El Emam, K., Ding, L. et al. A method for generating synthetic longitudinal health data. BMC Med Res Methodol 23, 67 (2023). https://doi.org/10.1186/s12874-023-01869-w
  • 1.3 Medical Data Anonymization:
        • Introduction: Medical data anonymization is crucial for protecting patient privacy while enabling the use of data for research and analysis. It involves removing personally identifiable information to prevent data from being traced back to individuals. In this topic, we will work on the anonymization of small datasets e.g. rare disease datasets.
        • Objectives:
                • Learn and study the basics of data privacy and data anonymization.
                • Extending our ArxPy tool functionality. Investigate the use of the tool for medical data anonymization.
                • Investigate the anonymization of small datasets such as our rare disease synthetic datasets.
                • Compare to available related work.
        • Resources:
                • Al-Dhamari et al, (2024). Synthetic datasets for open software development in rare disease research. Orphanet J Rare Dis 19, 265 (2024). https://doi.org/10.1186/s13023-024-03254-2
                • Walonoski et al. (2018). Synthea: An approach, method, and software mechanism for generating synthetic patients and the synthetic electronic health care record, Journal of the American
                • Medical Informatics Association, Volume 25, Issue 3, March 2018, Pages 230–238, https://doi.org/10.1093/jamia/ocx079
                • MacLeod, H., Abbott, J., Patil, S.: Small data privacy protection: An exploration of the utility of anonymized data of people with rare diseases. In: Proceedings of the 2017 Workshop on Interactive Systems in Healthcare (WISH’17) (2017)
                • Prasser et al (2020) Flexible Data Anonymization Using ARX — Current Status and Challenges Ahead. Software Pract Exper 2020;1–28. (Link)
                • https://github.com/iaBIH/synth-md
                • https://github.com/iaBIH/ArxPy
                • https://github.com/arx-deidentifier/arx

2. Trustworthy AI in Medical Imaging: Medical image analysis is crucial in modern health care as it enhances the accuracy and efficiency of diagnosing diseases. You will work with 3D multimodal image data using traditional and AI methods. We will investigate XAI (explanable AI) and Fair AI methods while solving real-world medical image analysis problems. Currently a student can work on one of these sub-topics:

  • 2.1 Medical Image Translation:
        • Introduction: Different medical image modalities, e.g. MRI, and CT, provide unique and complementary information about the human body, crucial for accurate diagnosis and treatment planning. Each modality captures distinct aspects of tissues and structures, enabling comprehensive assessments that are not possible with a single imaging technique. Image translation, which involves converting images from one modality to another using AI can significantly reduce the cost, time, and effort associated with medical imaging.
        • Objectives:
                • Learn and study the basics of AI, medical images, and image translation.
                • Using AI to convert a medical image to another e.g. modality translation and super-resolution.
                • Compare to available related work.
        • Resources:
                • Armanious et al. 2020, MedGAN: Medical image translation using GANs, Computerized Medical Imaging and Graphics, Volume 79, 2020, 101684, ISSN 0895-6111, https://doi.org/10.1016/j.compmedimag.2019.101684.
                • https://github.com/MedicalImageAnalysisTutorials/SlicerCochlea
                • https://github.com/MedicalImageAnalysisTutorials/SlicerCervicalSpine
                • https://trustworthyml.io/
  • 2.2 Medical Image Registration:
        • Introduction: Medical image registration is an important tool with many applications e.g. Image Fusion and Landmark detection. Solving the problem of medical image registration is challenging as there are many factors to consider e.g. the number of transform parameters and the similarity metrics.
        • Objectives:
                • Learn and study the basics of AI, medical images, and image registration.
                • Using AI to solve medical image registration problems.
                • Compare to available related work.
        • Resources:
                • Hering et al, (2022). Learn2Reg: comprehensive multi-task medical image registration challenge, dataset, and evaluation in the era of deep learning. IEEE Transactions on Medical Imaging. https://doi.org/10.1109/TMI.2022.3213983
                • https://github.com/MedicalImageAnalysisTutorials/SlicerCochlea
                • https://github.com/MedicalImageAnalysisTutorials/SlicerCervicalSpine
                • https://trustworthyml.io/
  • 2.3 Object and Landmark Detection in 3D Medical Images:
        • Introduction: Medical image registration is an important tool with many applications e.g. Image Fusion and Landmark detection. Solving the problem of medical image registration is challenging as there are many factors to consider e.g. the number of transform parameters and the similarity metrics.
        • Objectives:
                • Learn and study the basics of AI, medical images, and 3D object and landmarks detection.
                • Using AI to solve 3D object and landmarks detection in medical images
                • Compare to available related work.
        • Resources:
                • Takahashi et al. "Expandable YOLO: 3D Object Detection from RGB-D Images," 2020 21st International Conference on Research and Education in Mechatronics (REM), Cracow, Poland, 2020, pp. 1-5, doi: 10.1109/REM49740.2020.9313886.
                • https://github.com/MedicalImageAnalysisTutorials/SlicerCochlea
                • https://github.com/MedicalImageAnalysisTutorials/SlicerCervicalSpine
                • https://trustworthyml.io/

3. Practical Cellular Automata: Cellular automata have many applications in various scientific and engineering fields. For example, Rule 110 and the Game of Life are known to be Turing complete.  We will investigate the use of CA in two main topics i.e. security and simulation. Currently a student can work on one of these sub-topics:

  • 3.1 Cellular Automata Based Data Hiding:
        • Introduction: Ceullar Automata offer promising solutions for data hiding. The complexity and unpredictability of cellular automata make them ideal for data hiding e.g. image steganography and watermarking.
        • Objectives:
                • Learn and study the basics of Cellular Automata, Image Processing, and Data Hiding.
                • Investigate the use of CA in data hiding.
                • Compare to available related work.
        • Resources:
                • Azza A.A., Lian, S. Multi-secret image sharing based on elementary cellular automata with steganography. Multimed Tools Appl 79, 21241–21264 (2020). https://doi.org/10.1007/s11042-020-08823-8
                • Abu Dalhoum et al., 2012, , "Digital Image Scrambling Using 2D Cellular Automata," in IEEE MultiMedia, vol. 19, no. 4, pp. 28-36, Oct.-Dec. 2012, doi: 10.1109/MMUL.2011.54. 
                • https://plato.stanford.edu/entries/cellular-automata/
                • https://mathworld.wolfram.com/GameofLife.html
                • https://playgameoflife.com/
  • 3.2 Cellular Automata Based Simulation of Disease Spreading:
        • Introduction: Ceullar Automata offer promising solutions for simulation. The complexity and unpredictability of cellular automata make them ideal for simulation of complex phenomena.
        • Objectives:
                • Learn and study the basics of Cellular Automata, Image Processing, and Data Hiding.
                • Investigate the use of CA in data hiding.
                • Compare to available related work.
        • Resources:
                • Dai, J.; Zhai, C.; Ai, J.; Ma, J.; Wang, J.; Sun, W. Modeling the Spread of Epidemics Based on Cellular Automata. Processes 2021, 9, 55. https://doi.org/10.3390/pr9010055
                • Abu Dalhoum et al., 2012, , "Digital Image Scrambling Using 2D Cellular Automata," in IEEE MultiMedia, vol. 19, no. 4, pp. 28-36, Oct.-Dec. 2012, doi: 10.1109/MMUL.2011.54. 
                • https://plato.stanford.edu/entries/cellular-automata/
                • https://mathworld.wolfram.com/GameofLife.html
                • https://playgameoflife.com/