Automatic detection of cranial fractures in radiological images using a pattern classifier

Abstract

In this work, an automatic pattern classification system is presented, whose goal is detecting the presence or absence of fractures in cranial radiographic images. The basis for the proposal is an original coding technique, coupled with an emerging pattern classifier: the Gamma classifier. This proposal draws concepts from three areas of current scientific research: Mathematical Morphology, image histograms, and Alpha-Beta associative models. Also, an experimental study is presented, comparing the performance shown by the system to that exhibited by other pattern classifiers present in current scientific literature. The results obtained are competitive, reaching 94.23% of correct classification.