Search

Image mining is the application of computer-based techniques that extract and exploit information from large image sets to support human users in generating knowledge from these sources. This review focuses on biomedical applications of this technique, in particular automated imaging at the cellular level. Due to increasing automation and the availability of integrated instruments, biomedical users are becoming increasingly confronted with the problem of analyzing such data. Image database applications need to combine data management, image analysis and visual data mining. The main point of such a system is a software layer that represents objects within an image and the ability to use a large spectrum of quantitative and symbolic object features. Image analysis needs to be adapted to each particular experiment; therefore, ‘end user programming will be desired to make the technology more widely applicable.

Minimally invasive coronary artery bypass surgery needs careful planning, as the spatial range is much more restricted than in conventional approaches. We have developed the first steps towards a simulation and planning system for these procedures. Our approach is based on registering a virtual heart model to CT images of the patient. Such a model can embed empirical parameters about the intraoperative movement of elastic tissue to provide estimates for the motion range of target markers.

Virtual reality (VR) is one of the latest developments in cardiac three‐dimensional (3‐D) ultrasound. A VR heart model linked to 3‐D echocardiographic image datasets provides the observers spatial information regarding a 3‐D image dataset and prevents the “lost in space effect” in difficult and relevant coupled diseases when integrated into 3‐D reconstruction software. Standardized echocardiographic views can be selected within the integrated developed VR heart model, and this is the first step to automatic 3‐D computations with minimal operator interaction. VR heart models open exciting opportunities in the field of teaching echocardiographic cardiology, diagnosis, and examinable states.

It is important to provide a recovery operation for applications with a graphical user interface. A restricted linear undo mechanism can conveniently be implemented using object-oriented techniques. Although linear undo provides an arbitrarily long history, it is not possible to undo isolated commands from the history without undoing all following commands. Various undo models have been proposed to overcome this limitation, but they all ignore the problem that in graphical user interfaces a previous user action might not have a sensible interpretation in another state.Selective undointroduced here can undo isolated commands by copying them into the current state “if that is meaningful.”Furthermore, the semantics of selective undo are argued to be more natural for the user, because the mechanism only looks at the command to undo and the current state and does not depend on the history in between. The user interface for selective undo can also be implemented generically. Such a generic implementation is able to provide a consistent recovery mechanism in arbitrary applications.