Evaluating Visualisations and Automatic Warning Cues for Visual Search in Vascular Images

Visual search is a task that is performed in various application domains. The authors examine it in the domain of radiological analysis of 3D vascular images. They compare several major visualisations used in this domain, and study the possible benefits of automatic warning systems that highlight the sections that may contain visual targets and hence require the user’s attention. With help of a literature study, the authors present some theory about what result can be expected given the accuracy of a particular visual cue. They present the results of two experiments, in which they find that the Curved Planar Reformation visualisation, which presents a cross-section based on knowledge about the position of the blood vessel, is significantly more efficient than regular 3D visualisations, and that automatic warning systems that produce false alarms could work if they do not miss targets.