The Potential Use of Modified Seed-Based Region Growing Technique for Automatic Detection of Breast Microcalcifications and Tumour Areas

Breast cancer, a common female malignancy, does not show any symptoms in its early stage. Screening tests are therefore important to reduce the death rates. By far, mammography and ultrasound tests prove to be the tests for early detection of breast cancer. However, these tests have limitations such as blurriness, darkness and the existence of unwanted noise on the breast image, which can obscure breast tumours and microcalcifications; the salient features detected in cases of abnormality. Many image processing techniques have been introduced in order to detect the edges or segment these breast cancer morphologies including the seed based region growing (SBRG) algorithm. However, two parameters, namely the seed point and the threshold value of the conventional SBRG algorithm need to be determined manually. This paper attempts to automatically find these parameters by proposing the modified version of the SBRG algorithm. The proposed algorithm which is called the modified seed based region growing (MSBRG) algorithm has been tested on mammogram and ultrasound images to detect the breast tumour as well as microcalcification. The results show that the proposed method successfully detects and distinguish breast tumour and microcalcifications from the background and unwanted noises. The borders (edges) of regions found by the MSBRG algorithm are perfectly thin and connected. Hence, the size and shape of the regions will not be corrupted. These will certainly assist doctors in the breast cancer screening process.