Evaluation of multidimensional visualization techniques for medical patterns representation
Tomasz Rzeźniczak · Applied Computer Science · 2013
T here are many techniques of multidimensional data visualization. Part of them was built for a specific purpose, while another part is very general. In case of the former, the selec- tion of appropriate technique is straight forward, for the latter the selection is not always obvious. In certain simplification, selecting the most appropriated technique depends on the data which are visualized and the task that needs to be performed by the user over the visu- alization. This study is focused on evaluation of known visualization techniques for their applicability to a well-defined task and data set. The data set consist of medical patterns representing human diseases and their symptoms as an object-attribute data model. The task is to facilitate recognition of a disease by visualizing the reference medical patterns and the data of patient's condition. Human perceptual capabilities naturally predestine us to efficient receiving of visual s timuli. This is one of the reasons why information visualization is so widely used in vari- ous domains to facilitate data exploration, information understanding and communications. The field of information visualization grows very intensively and a lot of visualization tech- niques have been introduced. The common issue now is how to select the best technique for visualization of a particular data set for a particular task. Therefore, an even bigger focus in the area of visualization researches is put on evaluation of existing visualization techniques rather than designing new ones. The main goal of this study is to verify which of known visualization techniques can be applied and which is a best fit for visualization of objects from a medical patterns' data set. Beside the raw data set, a specific perceptual task type to be conducted over the visualiza- tion is also defined. The visualization should support recognition of objects from the given data set. It is expected that the visual representation of each medical pattern allows the ob- server to easily identify them, remember and extract information about its characteristics. The concept of leveraging visual representation for the mentioned perceptual task is a part of broader studies (30, 31), which introduce a visualization technique designed specifically for this purpose. At the same time no studies have been conducted yet to examine if any of the existing visualization techniques can be successfully applied to the given task type. This study aims to fill this gap. The process of medical diagnostic is an example of object analysis and recognition task, where a medical pattern is being detected and identified - among many known by a physi-