中文摘要：

提出了一种基于改进Snake模型的分割方法，用于数字人脑切片数据集中小脑组织的连续自动分割。在基本的Snake模型中添加了向心力、气球力，并采用自适应的能量约束项系数，根据相邻切片图像中小脑轮廓具有相似性，依次将单张切片的分割结果作为相邻切片分割的初始轮廓，进而实现整个数据集中小脑组织的连续自动分割。自动分割结果与专家手动分割结果的一致性较好，分割相似指数的平均值达到92．95％，最大为97．87％。结果表明该方法能够较为准确地从彩色人脑切片图像中提取出复杂的小脑组织，克服了现有方法对大量人工参与的依赖，并提高了分割的精度。

英文摘要：

A segmentation algorithm based on the improved Snake model is proposed, which is used in the continuous and automatic segmentation of cerebellum in the slice images of digital human brain. In this algorithm, centripetal force and balloon force are added into the energy function of basic Snake model, and the adaptive energy constraint coefficients are adopted. In the light of the similarity in the cerebellum contour of adjacent slice images, the segmen- ted result of the previous slice is used as the initial contour of the next one to realize the continuous and automatic segmentation of cerebellum in brain slice image sequence. The automatic segmentation results are consistent with the expert manually segmented ones. The average slice similarity index is 92.95% , and the maximum reaches 97.87%. The experimental results indicate that the proposed method can extract complex cerebellum in color brain slice images accurately, reduce the massive manual intervention involved in existing segmentation methods and improve segmentation accuracy.