Honours Projects 2009

Projects supervised by Jinman Kim

Biomedical Information Technology
There have been tremendous advances in biomedical imaging hardware and software technologies. Modern scanners such as the combined Positron Emission Tomography and Computed Tomography (PET-CT), that sequentially acquires high-resolution imaging (3D volumetric data) of the human function (PET) and anatomy (CT), have dramatically changed the way biomedical image data are used and accessed – in diagnosis, distribution and management. We have a strong relationship with our partners in the Royal Prince Alfred (RPA) hospital which ensures that our research is clinically relevant and have access to the state-of-the-art hospital facilities. For further details, please refer to our biomedical and multimedia information technology (BMIT) research group’s website.

Video-based Representation of Large Medical Data
Supervisor: Dr Jinman Kim, Dr Peng Gong, and Prof Dagan Feng
Digital distribution of medical images is introducing improved patient care and cost benefits and will continuously grow in importance. The use of modern streaming and quality-of-service algorithms relies on video formats such as FLV, MP4 and 3GP, or other stream-able formats such as JP3D. Medical image sequence data, for which compression quality is critical, must be transcoded into a high-fidelity video stream before such techniques may be applied. This project looks at building an automatic transcoder to convert medical image sequences into video files, and creating a simple medical image viewer to display such video-sequences. This project will extend on our biomedical mobile imaging system that is currently in development in the Royal Prince Alfred (RPA) hospital.

Secure Medical Record Forwarding via Smartphone Messaging Protocols
Supervisors: Dr Jinman Kim, Dr Peng Gong, and Prof Dagan Feng
Specialised medical staff needs to be able to communicate cases and findings at the push of a button, so as to optimally pool their collective knowledge and experience. This project aims to create a prototype messaging service which allows the transmission of facts and image data directly from a medical database system to a physician’s Smartphone. The project will involve integration into a hospital database interface, and web-based mobile representations of this interface, and require collaborative work with other students and medical staff. This project will extend on our biomedical mobile imaging system that is currently in development in the Royal Prince Alfred (RPA) hospital.

Multi-modal Image Fusion Visualization for Biomedical images
Supervisors: Dr. Jinman Kim and Prof. David Feng
Next-generation of medical imaging scanners are introducing new diagnostic capabilities and improved patient care. However, these modern scanners are restricted from its full usage capacity due to the lack of interactive visualization for diagnosis needs, where slice-by-slice approach is currently the norm. The massive number of images (1000’s) and the complex inter-relations between the functional (PET) and anatomical (CT) images will mean that access to and assimilation of critical data within these images, by the reader and end-user (e.g. neurosurgeon, cardiothoracic surgeon) will become a major problem. Thus a new method of visualization is essential. In this project, we will develop a new data fusion algorithm in order to maximize the visual information in an interactive volume rendering of dual-modal PET-CT images. Image fusion is the process of creating new information from combining multiple image sources (data reduction) for optimal visualization. This project will build up on our medical visualization system stationed in the Royal Prince Alfred (RPA) hospital.

Anatomical Image-based Estimation of Accurate Parametric Measures for Functional Imaging
Supervisors: Dr. Jinman Kim, Dr. Lingfeng Wen, and Prof. David Feng
Parametric images like the standardized uptake value (SUV) measure, is widely used in clinical oncology to quantitatively assess the malignancy of lesions in the staging of cancer and evaluation of treatment response. However body weight, used in the calculation of SUV only provides a semi-quantitative estimate of glucose metabolism in the whole body. This project will investigate a new approach to use imaging information in co-registered PET-CT images to provide more accurate parametric measure to aid the diagnosis of lesions. Anatomical information available in CT data will be used to derive improved more accurate measures as well as improved non-invasive estimation. New framework will be developed at the base of our already developed image processing software package. This project will present an opportunity to access the state-of-the-art medical imaging technology and to work with staffs at the Royal Prince Alfred (RPA) hospital.

Computer Aided Diagnosis of Dual-Modal PET-CT data for cancer
Supervisors: Dr. Jinman Kim, Dr. Lingfeng Wen, and Prof. David Feng
Rapid advances in medical image scanners have led to improved image quality for more reliable diagnosis by multi-fold increasing spatial and temporal resolutions. Modern scanner systems are capable of acquiring 1000s of image slices of the whole-body in a short period as compared with a few slices can be obtained many years ago. The interpretation of these images, however, is still almost exclusively the work of end-users e.g., medical doctors and specialists. This has created a huge demand for computer-aided diagnosis (CAD), a computer system that is used to aid in image interpretation and diagnosis, by providing a second option using information technology in biomedicine. One of the most important need for CAD is the ability to automatically detect lesion candidates and to identify whether this lesion is benign (non-fatal) or malignant (fatal). A CAD system will be developed for the identification of lesions in whole body PET-CT data by using the features derived from the images. Our previously developed segmentation algorithm will be improved and used to extract lesions that will be used to train and build up a database of lesion patterns for CAD system. This project will present an opportunity to access the state-of-the-art medical imaging technology and to work with staffs at the Royal Prince Alfred (RPA) hospital.

Temporal Change Detection of Multi-Modal PET-CT data series
Supervisors: Dr. Jinman Kim, Dr. Lingfeng Wen, Dr. Xia Yong, and Prof. David Feng
Serial PET-CT scans to assess treatment response or lack of response is an important area of research in malignant disease. Due to the large number of sites of disease, visual assessment of responses is problematic and time consuming. An automated method of response assessment and feasible visualization would be advantageous however the large image dimensions and different degrees of FDG uptake in the full 3D data even for same patient pose major barriers. This project will investigate an automated approach for fast change detection in serial whole body PET-CT data by combining two image processing methodologies of image subtraction and large-data clustering and investigating new feasible visualization of those changes.. We will innovate in the use of smart algorithms that make advantages of the co-aligned functional and anatomical information available in dual-modal PET-CT data and our already developed visualization package. This project will present an opportunity to access the state-of-the-art medical imaging technology and to work with staffs at the Royal Prince Alfred (RPA) hospital.

Feature Extraction of Biomedical Images for Content-based Image Retrieval
Supervisors: Dr. Jinman Kim, Dr. Tom Cai, and Prof. David Feng
With the ever-growing image databases in hospitals and other clinical environments, the keyword-based searching and storage solutions currently employed are restricting the full accessibility of the information available in these images. There are increasing demands in image retrieval capabilities, such as the need to retrieve spatially-relevant (visually) images for cancer patients, i.e., evidence-based diagnosis, medical education, and biomedical research.

This project will focus on automated extraction of features that will enable content-based retrieval. Features to be extracted will include textures, anatomical structures, functional structures, image histogram, etc. Spatial features from co-registered functional and anatomical PET-CT images will be segmented and the relationships between these features will be used to construct invariant graph representations. These graphs will then be used to search for relevant images that share relevant spatial relationships among the features. This project will extend on already established content-based retrieval system developed in-house, and present the possibility to work with staff at the Royal Prince Alfred (RPA) hospital.

An Image-Query Interface for Content-Based Image Retrieval of Biomedical Images
Supervisors: Dr. Jinman Kim, Dr. Tom Cai, and Prof. David Feng
Retrieval of patient data based on PET-CT imaging features is a novel complement to text-based search methods; furthermore, such accumulated previous knowledge may aid clinical decision-making and treatment planning. There are increasing demands in image retrieval capabilities, such as the need to retrieve spatially-relevant (visually) images for cancer patients, i.e., evidence-based diagnosis, medical education, and biomedical research.

This project will involve the design and development of a touch-screen user interface to allow query graphs to be drawn using gesture for retrieval. Because of the complexities introduced in multi-modal imaging, the ability to ‘draw’ the query image can be very useful. This project will extend on already established content-based retrieval system developed in-house, and present the possibility to work with staff at the Royal Prince Alfred (RPA) hospital.