The aim of this project is to apply image processing methods to X-ray–guided needle insertion, with a focus on improving fluoroscopy image quality to support accurate needle positioning during minimally invasive procedures. The project will concentrate on the real-time analysis of X-ray fluoroscopy images, including needle detection and tracking, basic tumour localisation, and image-based guidance. Machine learning–based methods will be used to enhance X-ray image quality and provide robust visual guidance during needle insertion.
The proposed techniques will first be validated using in vitro models, such as gelatin or agar gel phantoms, to track needle motion and evaluate insertion accuracy. In addition, the project will consider how the needle tip can effectively cover a target region within soft tissue, representing the tumour, in order to support complete “dose painting” with microparticles. The performance of the methods will be evaluated through experimental studies and validation using medical imaging data.