Pixel-Based Morphological Technique for Breast Tumour Detection

Breast region segmentation is the process of splitting mammogram image into breast region and background to focus and limit the search for abnormality on the breast region without the effect of the background on the results. In addition, performance of existing Computer Aided Detection (CAD) systems for detection of malignant tumours in breast tissue have been limited by the methods of segmentation. Image segmentation is a multi-objective problem where multiple criteria must be considered for extraction of breast region. The developed segmentation technique in this paper considered intensity of pixel for image binarization (using Otsu thresholding) and shape for image boundary refinement (using mathematical morphological processes), to detect exact location of tumour in breast tissue. The developed technique was evaluated using Kappa agreement scale (Hit, Miss and Over-hit). A moderate value of 0.59 in Kappa agreement scale was achieved for the segmentation. The Two-stage Segmentation Technique is efficient to extract the locations of breast tumour with low level of false positive. Key words— Segmentation, pixel, morphology, tumour, Hit, Kappa, benign and malignant.