Intensity versus texture for medical image search and retrival
Devrim Ünay, Ahmet Ekin · 2008
The ever-increasing number of digital images in the medical domain, has amplified the need for automated search and retrieval tools. Furthermore, medical experts generally focus on specific anatomical structures to identify the cause of a pathology. For such cases, automated tools that can retrieve relevant slice(s) from a patient's image volume can assist the expert in diagnosis. Accordingly, in this paper we introduce a new search and retrieval work for finding relevant slices in brain MR (magnetic resonance) volumes. The features explored in this framework are based on intensity, texture, and their extended versions complemented with spatial context. Experiments on real data revealed that texture information outperformed its intensity counterpart, incorporating spatial context in the features substantially improved the accuracy, and finally texture features with spatial context provided fast and highly accurate retrieval of relevant slices.