Abstract:

In order to overcome the instability and uncertainty of the volume segmentation of medical images via the conventional active contour model, this paper proposes an improved active contour model based on the pertinence of gray scale and space position between medical series images. In the proposed model, an internal energy function is constructed according to the distance between the left neighbor midpoint and the right adjoining midpoint, which is limited by restricting the contour length. Then, an external energy function is constructed according to the local gra- dient information of image as well as the overall information and its pertinence in the local region between series images. The weight parameters are automatically regulated according to the ratio of internal energy to external energy. Experimental results show that the proposed algorithm can detect some corner and concave points and can prevent the object edge from converging to noise and pseudo edge points, thus resulting in good volume segmentation effect.

Key words: active contour model, volume segmentation, energy function, series image, pertinence