Fast Chan-Vese without edges and connected component analysis for kidney segmentation in MRI images

In this paper, fast Chan-Vese (CV) with simple shape model and connected component analysis (CCA) is proposed to segment kidneys in Magnetic Resonance Imaging (MRI). The performance of the CV model is improved by considering the sign of the level set in single layer (gray level) intensity image taking only the fitness term of the energy functional. Connected component analysis is performed on the mask obtained from final level set to earmark the possible candidate kidney. Once the candidate objects are identified, shape model is imposed to obtain the kidney. For the same size and placement of initial contour on the MRI image the fast CV has shown superior performance to that of the original CV model. In addition the paper also compares Otsu´s thresholding algorithm with fast CV method. It then established that both Otsu´s global thresholding and the proposed fast Chan-Vese model with single level set method achieve comparable performance for kidney segmentation. Further, the paper shows that Otsu´s adaptive thresholding model is superior to fast CV model in terms of speed.