A hybrid segmentation approach for brain tumor extraction and detection

Abstract

Nowadays, medical image processing and particularly MRI images is the one of the most challenging field and emerging to help specialists in their diagnostics. In this context and to detect automatically suspicious regions or tumors, this paper presents a new approach called hybrid segmentation inspired by both mathematical morphology operators and morphological watershed segmentation. Our approach's advantage comes from the complementarity between these two approaches. The morphological operators extract roughly the tumor region and eventually can affect healthy structures while the watershed method provides details of various brain's structures and therefore the fusion of these two approaches improves significantly the segmentation and the extraction of the tumor zone.