Honours Projects 2008

Projects supervised by Xiuying Wang and David Feng

With rapid advance in imaging technologies, huge volumes of images can be easily acquired from diverse imaging sensors (such as infrared, laser, SAR), modalities (such as MR, PET, CT) and sensor networks. Effective and smart use of complementary information embedded in multimodality imaging data is crucial for applications in remote sensing, multimedia areas, and life science and healthcare. Image registration is an effective mechanism for maximizing the information embedded in image datasets. By establishing spatial correspondence among the multiple datasets, medical image registration enables integrating and fully utilizing heterogeneous image to facilitate accurate diagnosis and management of patients with a variety of diseases.

Automatic Registration for 3D Whole-body Images from Combined PET/CT Scanners (18cp)
The combined PET/CT system, which provides hardware registration between functional information from PET and anatomical structure from CT, has been widely accepted in clinical practice. However, misregistration of PET/CT volumes may be introduced due to patient’s motions, disease progression or treatment intervention. This project is to provide an optimal solution to registration of datasets obtained from the combined PET/CT scanners with an automatic strategy. The project also includes the investigation of the influence of different optimization strategies on medical image registration. Through this project, students can further their knowledge of image processing and keep touch with the state-of-art biomedical image registration techniques. The project will offer opportunity to work with clinical staffs at the Royal Prince Alfred Hospital and can be extended as a PH.D topic.

Deformable Registration for Temporal Lung Volumes (18cp)
Complex deformations of lungs due to involuntary motions of respiratory and cardiac processes pose particular challenges for generating accurate registration with improved performance and accelerated computation. This project is to address these issues by including local structural information to achieve more accurate and faster deformable registration for datasets obtained from modern advanced imaging systems. Automatic organ extraction and segmentation will be involved in this project. The project will offer opportunity to work with clinical staffs at the Royal Prince Alfred Hospital and can be extended as a PH.D topic.