Inter-patient analysis of tomographic data

Abstract

The paper considers the computer-based support for the localization of pathological tissue within tomographic data. The subject of the approach is the inter-patient analysis of brain tissue types such as tumor, CSF, white matter, grey matter, bone, fat tissue and background. The class tumor hereby represents the superset of pathological tissue. The analysis pipeline of the presented approach contains feature extraction, classification, two-step texture analysis and morphological postprocessing. Furthermore the paper reports results that have been reached on the different steps of the pipeline.