A Brain Tissue Segmentation Approach Integrating Fuzzy Information into Level Set Method

In this paper, we propose a new segmentation approach based on level set techniques to segment the brain MR images. We adopt a new binary regional term based on the fuzzy information of the image in the new algorithm, which can inflate or contract the evolving curves automatically without predefined the evolving directions during the initialization phase. The algorithm can segment brain tissues from the different modalities MR images with the same parameters. We compare the performance of the new algorithm with the primary algorithm by simulated experiments. We also explore the influence of the parameter setting and the binary processing of regional term to the algorithm by experiments and statistical analysis. The quantitative and qualitative analysis show that the new algorithm provides more accurate segmentation results with good robustness, and is less sensitive to parameter setting. Furthermore, the binary processing of the regional term greatly decreases the number of iterations; namely, it makes convergence of the new algorithm more quickly.