Automatic exclusion of skin border regions from breast MRI using proposed combined approach

In this paper, an automatic combined approach is presented to exclude skin border regions from Breast MRI images. This procedure is vital as pre-processing process in breast tumour segmentation systems. In order to detect and remove skin border regions, the proposed approach integrates two methods in two main stages. These stages are: edge detection using Level Set Active Contour method. Followed by, skin border removal using Morphological thinning method. The results showed high performance for edge detection stage with average of sensitivity 0.81 and specificity 0.94. On the other hand, the final skin border removal stage is also scored high performance with average of sensitivity 0.86 and specificity 0.97.