Current projects

Developing Deep Learning models for the automated diagnosis of stroke utilising imaging data

Our team is leveraging state-of-the-art techniques in deep learning and medical imaging analysis to revolutionize stroke diagnosis. With expertise in convolutional neural networks (CNNs), recurrent neural networks (RNNs), and attention mechanisms, we aim to develop highly accurate and efficient models capable of analysing various imaging modalities, including MRI, CT, and angiography scans. By incorporating advanced image segmentation algorithms and transfer learning techniques, we optimize model performance and generalizability across different patient populations and imaging protocols.

Moreover, our team prioritises the interpretability and explainability of our deep learning models, ensuring that clinicians can trust and understand the diagnostic decisions made by the algorithms. Through rigorous validation studies and collaboration with stroke clinician experts, we are validating the performance of our models against gold-standard diagnostic criteria and clinical outcomes, ultimately enhancing diagnostic accuracy, efficiency, and patient care in managing stroke.

Developing multimodal large language models for monitoring brain functions and disorders utilizing EEG, fMRI, MRI, and questionnaire (e.g., clinical report) information.

We are working to develop multimodal large language models for monitoring brain disorders by utilizing a combination of EEG, fMRI, MRI, and questionnaire data, such as clinical reports. This involves integrating various types of information related to brain function and disorders into a unified framework. The project will leverage advanced natural language processing techniques to analyse textual data from clinical reports and questionnaire responses, alongside signal processing and image analysis methods for EEG, fMRI, and MRI data. The goal is to create AI systems capable of detecting patterns, abnormalities, and biomarkers associated with brain disorders across multiple modalities, ultimately enhancing diagnostic accuracy and treatment planning in clinical settings.