Towards an open-source, equipment-agnostic framework for automated welfare monitoring in the home cage

Mice are the most widely used model organism in all of science. In 2018, 60% of all experimental procedures performed in the UK were carried out using mice, amounting to more than 2.5 million procedures in total. While the vast majority (~90%) of procedures were classified as being of “mild” or “moderate” severity, there is an ongoing unmet need for technological solutions to the problem of long-term, continuous welfare monitoring to determine humane experimental endpoints in an unbiased way.

Your project will focus on the development of computer vision/machine learning techniques to monitor the behaviour of mice in the home cage, with a view towards developing a general-purpose open-source framework for the animal research community. Anomaly detection is a technique whereby rare or abnormal events are classified as deviations from what is otherwise expected, based on an existing body of “normal” data. Unsupervised deep learning techniques, such as generative adversarial networks (GANs) and variational autoencoders (VAEs), offer great promise for the detection of anomalies in both image and video data. You will develop techniques along these lines using hundreds of hours of video footage acquired of mice via an infrared camera. This project will provide you with ample opportunity to explore a wide range of techniques for the detection and localisation of abnormal behavioural events using real-world data.

The University of Lincoln is seeking to appoint a motivated, inquisitive PhD candidate to join a diverse team of researchers in the School of Computer Science. You will have a masters (ideally) or bachelor’s degree (2:1 or above, with honours) in computer science, engineering, bioinformatics, or a related discipline. You should also demonstrate one or more of the following in your application:

- Working knowledge of Python, MATLAB, or a similar high-level programming language
- Familiarity with classical image/video processing and computer vision techniques
- Experience in using one or more deep learning libraries (e.g., Keras, TensorFlow, Torch)
- Contributions to open-source software projects related to imaging/vision/machine learning
- 1+ years’ experience in machine learning or (bio)statistics, either in academia or industry

The Laboratory of Vision Engineering (LoVE), in the School of Computer Science, is a team of internationally-recognised computer vision researchers. Our research focuses on several application areas including healthcare, security, environment, and animal welfare. You will join our diverse team of academics, postdocs and postgraduate students led by Prof Nigel Allinson MBE (Distinguished Chair of Image Engineering, College Director of Research). Your project will be undertaken under the supervision of Dr James Brown (Senior Lecturer/Assistant Professor in Imaging/Vision) and Prof Xujiong Ye (Professor of Medical Imaging & Computer Vision, School Director of Research). You will also have the opportunity to collaborate and visit with leading researchers in mammalian genetics at the Mary Lyon Centre, MRC Harwell Institute (Oxfordshire). They will provide invaluable insight, guidance, and data in support of this project for its duration.

The University of Lincoln is committed to fostering a culture and environment where individual differences are appreciated and respected, ensuring equitable and fair treatment for all. Applicants will be considered without regard for age, disability, gender, marital status, pregnancy/maternity, race, religion, sex, or sexual orientation.