Comparison of supervised and unsupervised classifications in the detection of hepatic metastases
An improved genetic K-means clustering algorithm is proposed and is applied to image segmentation. According to the characteristics of the image, the feature vector of the pixel is properly chosen and the weight factors of the feature vector are adjusted, which enhances the segmentation precision. The selection of conventional genetic algorithm and the modification of mutation operations improve the speed of convergence. Computing time is reduced due to combining the membership matrix with the coding of chromosomes skillfully. The results of the experiments demonstrate that in the image segmentation the proposed algorithm is better than traditional genetic K-means algorithm.