A new method for mammographic mass segmentation based on parametric active contour model

Image segmentation is an important task in the analysis of mammograms and it is challenging because the masses are low contrast with ambiguous margins. The classical Vector Field Convolution (VFC) method has demonstrated its merits in image segmentation; however it is difficult to capture the ambiguous object boundaries. In this work, a new snake model is proposed to solve this problem, this proposed method combines the morphological filter with the laplacian operator. A new external force is calculated by convolving the edge map with the user-defined vector field kernel. The proposed algorithm is tested on the Mammographic Image Analysis Society (MIAS) database. The experimental results suggest that the proposed method can effectively locate ambiguous margins of the masses. Compared with gradient vector flow (GVF) snake and classical VFC snake, the proposed method shows its advantages, including the reduced computational cost and better performance.