Project offerings for Semester 1 2014
Standard tex michael kors bag t based descriptive fields are insufficient for describing rich image properties, and hence more intelligent methods to represent image features are desired. Currently many different ways of feature description and similarity measure have been proposed in the literature, and the bag of feature (BOF) approach and its numerous variations have become widely popular. This project aims to study the various existing BOF based approaches, implement them for medical image retrieval on large scale imaging databases, and compare their performances in terms of retrieval precision and recall.
Multi View Visual Feature Encoding for Image Pattern Classification (18 cp)
Image pattern classification has a wide variety of applications, such as differentiation of disease patterns and detection of interest objects. The classification performance is largely dependent on the descriptiveness and discrminativeness of feature descriptors. Consequently, how to best encode the complex visual features into feature descriptors is crucial. Currently many different ways of image feature extraction have been proposed in the literature, and how to combine the multi view features to achieve effective classification is of particular interest. This project aims to study the various techniques of image feature extraction and multi view feature combination, to develop a new methodology for multi view visual feature encoding, and to evaluate the performance for classification on multiple medical imaging databases.
Projects supervised by Jinman Kim, Simon Poon and Mohamed Khadra (Nepean Hospital)
Nepean Telehealth Technology Center (NTTC) (18cp)
The field of Telehealth the use of telecommunications technology to conduct medical diagnosis, treatment and monitoring, has developed far beyond its initi michael kors bag al role of providing healthcare consultations to people living in remote communities via video conferencing. We are now conducting research to look at technologies such as remotely controlled robotic surgery, smart medical homes, automated patient monitoring, and using mobile technologies to provide community based monitoring, medical image sharing, consultation and treatment by healthcare practitioners.
These Telehealth researches are conducted at the Nepean Telehealth Technology Centre (NTTC). It was established in 2013, with funding from the state government, with the objective of introducing new Telehealth technologies and innovations to the Nepean hospital environment. Nepean is a strategic location to conduct Telehealth due to its geographical location that covers large rural areas where Teleheatlh can make significant benefits. A unique multi disciplinary partnership between the Nepean Hospital and the institute of Biomedical Engineering and Technology (iBMET) at the University of Sydney, was established to innovate in new Telehealth technologies. This partnership enables IT research with direct applications to the clinical environment and sharing / translation of technologies. In this project, we will develop a surgical simulat michael kors bag ion training tool, akin to the Da Vinci robot, but to be low cost and readily accessible. This project will expand upon our research into designing user interfaces for patients and elderly to track and monitor their health care.
We further seek students who can bring their own experiences and inter michael kors bag ests in Telehealth to create new projects or to customise one of the projects above. Students will join a team of research student and will have the opportunity to work at the clinical environment and with clinical staffs / students (Nepean Hospital clinical school); as well as to join the iBMET (Level 5 West, School of IT Building).
For more information please visit the NTTC and iBMET websites.
Project supervised by Karsten Klein
Interactive 3D Data Visualization (18cp)
Recent advancement in technology makes affordable and easy to use 3D equipment available. In particular the new Oculus Rift device is getting much attention not only among developers, but also in the media, and a lot of development work is done to make VR applications ready for use with the Rift.
This project aims at exploring the potential of such devices for data visualization, which includes creating and testing visualizations as well as designing the corresponding interactions with the visualization interface. The project results should constitute a first
step towards the future goal of using the Rift as a tool within a visual analytics approach for
complex data sets.
Visual Analytics helps to make sense out of huge and complex data sets by combining interactive visualizations with automated analysis methods. This approach allows the human user, often an expert in his field, to gain insight and to generate hypotheses on the structure of the data through analytical reasoning.
In today’s Big Data era, analyzing such data sets is crucial in a large variety of application areas, including for example social network analysis, the financial sector, security, and biology. Often, relational information like social network friendships, stock market trader relations, or protein interaction, is modeled and visualized as a graph for analysis purposes.
An important part of this project will thus be to develop graph visualizations and graph navigation for use on the Rift device. Data from real world projects will be provided to the student.
Research task: Create visual interfaces that allow to represent complex data sets in 3D using the Oculus Rift device.
Skills and requirements: Good programming skills are needed, a willingness to be creative, knowledge of either graph or gaming visualization or image rendering, plus corresponding software and development tools, and an interest in learning about visualization methods.
The student will learn how to create interactive visual interfaces and how to employ 3D visualization hardware.
Visual Analysis for Graph Set Properties (12/18cp)
Visual analysis can foster efficient analysis of complex data sets, and facilitate human reasoning, such that patterns in the data can be discovered that would remain hidden otherwise. biological molecules with their physical and biochemical properties, or companies with performance indicators over time.
The goal of this project is to create a visualization for such a data collection, a set of graph properties that was created based on several publicly available graph benchmark sets. potential clusters, the distribution of property values, and differences and overlaps between the different benchmark sets. The result might either use a static, map based approach or allow interactive exploration of the data via a GUI.
Students will gain good knowledge in Information Visualization and learn how to create pictures of abstract data.
The outcome might have the potential to be used in a long term research project and to be made publicly available.