“…The results of the proposed method are compared with 11 kernel graph cut methods: RBF, 11 entropy-based, 19 Krawtchouk, 24 L2S, 37 LoGRSF, 38 LocalPrefitting, 39 AMOE, 28 edge reg, 32 curvature reg, 33 spatial reg, 34 and customized RBF. 27 It should be mentioned that, to make a fair comparison, the coefficient of the regularization term in the RBF and entropy algorithm is considered to be the middle limit and is equal to 0.5. To evaluate the performance of the above segmentation methods, six criteria are adopted: Dice similarity, Jaccard similarity, 40 accuracy, 41 peak-signal-to noise ratio (PSNR), 42 mean sum of squared distance (MSSD), 43 and elapsed time.…”