N. Lassouaoui and L. Hamami (Algeria)
Mathematical morphology, edge detection, segmentation on homogenous areas, medicalimage, brain image.
In this paper, we present algorithms which permit to
extract tumors zones in brain image. The goal is to
extract the abnormal zones with respect of theirs
characteristics as: size, shape and position. Then we
present to the doctor a simple image of abnormal zone
that is easy to analysis and to diagnosis.
For it, we used an algorithm of edge detection for
delimit this areas; moreover, we have used other
algorithm which permits to extract the areas where the
tumors exist. These algorithms are based on
mathematical morphology operators. The results
obtained are satisfactory, the edge detection respects the
characteristics and the morphology of original sick
brain image, whereas with the algorithm of
segmentation in homogeneous areas, we could make the
extraction of the sick part of the brain, with the respect
of all characteristics in a reasonable time allowing to
think of real-time operation.