Blood film morphology
Visual assessment of blood films by an expert is time-consuming and subjective. This multi-disciplinary theme aims to develop a new machine-learning tool with the minimal required hardware for the automatic analysis of blood samples of heterogeneous populations.
The first phase of the theme will focus on developing machine learning models (Vision Transformer (ViT), including Swin Transformers) for detecting leukaemia in blood films gathered from an east London population.
The second phase of the project will extend the use of the models on blood films to detect other diseases (malaria, anaemia, thrombocytopenia, sepsis), and to validate their accuracy on different populations (including children and different geographies).