Applications of 3D Sensing to Novel ProblemsBob Fisher
School of Informatics, Edinburgh University
The availability of the Kinect has opened the door to easy 3D applications, mainly tailored to people and indoor scenes. This talk will present two novel applications enabled by high precision 3D sensors. The first is an exploration of the benefits of 3D to skin cancer segmentation and diagnosis, using a high resolution registered depth and color image. The second application concerns live bat tracking and shape analysis, using a 500 frame per second range sensor. Both use novel high precision sensors.
Download PDF [678 KB]
Quantitative Imaging Biomarkers of Cardiovascular and Neurodegenerative DiseaseWiro Niessen
Biomedical Imaging Group Rotterdam
Erasmus MC Rotterdam, Delft University of Technology
In this presentation, the development and validation of quantitative image analysis techniques to improve diagnosis, therapy planning and therapy monitoring in cardiovascular and neurodegenerative diseases will be discussed.
In the management of cardiovascular and neurodegenerative disease, advances in imaging devices have drastically increased our capabilities to (non-invasively) study both anatomy and function. With these advances, the sheer size, complexity, and heterogeneity of imaging data available for biomedical research and clinical practice have increased enormously. There is currently a lack of adequate and validated image processing techniques to analyze these data. Within the cardiovascular domain, this presentation will show advances in methods to relate morphological and anatomical characteristics of cardiovascular disease to cardiac function. Additionally, techniques for the quantification of atherosclerotic disease from non-invasive imaging techniques, for improved diagnosis and prognosis will be presented. Within the neurodegenerative domain, we will show how large scale analysis of neuro imaging data, both from population imaging studies and clinical studies, can lead to novel methods for early detection, and differential diagnosis of neurodegenerative disease.
Download PDF [34.5 MB]
Hyperspectral Image ProcessingBioucas Dias
Instituto Superior Técnico
Hyperspectral instruments acquire electromagnetic energy scattered within their instantaneous field view in hundreds or thousands of spectral channels with much higher spectral resolution than their multispectral antecessors. This enhanced spectral resolution has opened the door to countless applications where the ultimate objective is the identification of materials, or states of materials, at each pixel based on the respective spectral properties.
However, the large dimensionality of the hyperspectral data cubes (i.e., two spatial dimensions plus the spectral dimension) together with the degradation mechanisms usually present in the acquisition process (e.g., radiometric distortions, spectral mixing, blurring, and noise) call for powerful image processing methods. In this talk, I will present an overview of relevant inverse problems in hyperspectral imaging, namely denoising, unmixing, debluring, and superesolution. For each topic, I will summarize the mathematical problem involved, give relevant pointers to state-of-the-art algorithms to address these problems, and illustrate experimentally the potentialities and limitations of these algorithms.
Download PDF [2.92 MB]
Exploratory Data Analysis in Dynamic EnvironmentsRudolf Kruse
Computational Intelligence Research Group
Department of Computer Science, University of Magdeburg
Temporal databases have become an integral part in many economic fields. One challenge is to exploit the dynamic nature of the data to solve specific optimization problems in the underlying domains. In this talk, I will present several challenging real-world applications where explorative analyses of temporal databases are de- manding. I will give examples of dealing with different models to incorporate the temporal aspects. Firstly, I will focus on the temporal analysis of classical time series generated in the field of automobile development and manufacturing. Here, one task is to detect and predict changes during the manufacturing process by exploratively analyzing certain interestingness measures of dynamic association rules. The purpose of another example in the same field is to dynamically model automo- bile drivers based on their driving behavior. Secondly, I will discuss a new trend in explorative data analysis that deals with complex dynamic networks as they occur in social communities or biological networks. In the form field, dynamic graphs are exploratively analyzed to track temporal activities of users. This kind of analysis has been applied to several datasets, e.g. spatio-temporal data from the Second Life community. In the latter field, static clinical and highly dynamical neuroimaging data from patients suffering from vision loss are analyzed in an explorative way to support medical treatment. Here, temporal models of observed dynamic graphs are used to describe coherences with clinical variables. All methods offer intuitive ways to guide the user through the process of data and model inspection and assist in drawing conclusions with the help of meaningful graphical representations.
Download PDF [2.07 MB]
ICIAR - International Conference on Image Analysis and Recognition
Send your comments to firstname.lastname@example.org