Improved live-wire algorithm for interactive medical image segmentation

Lung extract is very important for computer-aided diagnosis, and it provides reliable diagnostic data for doctors. This paper presents an interactive lung segmentation method. Firstly, using optimized Gabor odd filter to filter chest X-ray image, and we can get a map of edge response energy. Then, using the edge response energy values to construct livewire cost function, which can segment chest image. Test shows that this algorithm could correctly-distinguish strong and weak edge, and extract the lung's contour quickly and effectively. Compared with the traditional algorithm, it reduces the number of human-computer interactions, which is more robust and efficient.