Positions: Student researcher
Created: 2023-10-23 Deadline: 2023-12-31
Location: United Kingdom
Most machine learning methodologies require large datasets with high-quality labelling to effectively train and validate the developed models. However, as medical imaging repositories used for population health studies, such as the UK Biobank, continue to grow, the conventional manual annotation approaches by experts become impractical. On the other hand, the labelling process can be automated by extracting annotations from associated metadata or reports. In such scenarios, automated labelling can introduce errors due to ambiguities in the reports, resulting in large but noisy-labelled (or even mislabelled) datasets. Consequently, this can lead to poorer generalization or replicate human biases present in the data. The primary objective of this project is to investigate machine learning strategies that can guide optimal annotation techniques to achieve high accuracy while mitigating biases in the developed model.
The student will be expected to attend relevant seminars within the department and those relevant in the wider University. Subject-specific training will be received through our group’s weekly supervision meetings. Students will also attend external scientific conferences where they will be expected to present the research findings.
The ideal student would have: a degree in computer science, statistics, engineering or a related discipline strong programming skills (preferable Python, or Matlab/C++ and willing to learn Python) experience or interest in machine learning (Deep Learning) and medical image analysis experience or enthusiasm to work on clinically relevant problems.
Contact: Bartek Papiez (bartlomiej.papiez [ at ] bdi.ox.ac.uk)
The Oxford Biomedical Image Analysis (BioMedIA) cluster is an academic group of faculty, postdoctoral researchers, software engineers, support staff and research students that develop medical imaging and image analysis algorithms and tools that aim to improve image-based diagnostics, therapies and monitoring technologies in hospitals and primary care, and for both western world and global health care settings. The breadth of our interests span all major clinical imaging modalities (particularly magnetic resonance imaging, ultrasound imaging, endoscopy imaging, histopathology), multi-modal imaging (imaging and audio, imaging and gaze tracking, imaging and electrocardiogram) and microscopy. We conduct inter-disciplinary translational research with clinical partners in Oxford and elsewhere in the UK and overseas in clinical domains of application ranging from fetal development, to oncology, respiratory medicine, gastroenterology, neurology and cardiovascular medicine.
See lab's page