Abstract: In order to segment the extern contours of hip joint correctly and effectively, a kind of extraction method for extern contour of proximal femur is presented based on the best segmentation curve. The proposed method can find a closed segmentation curve which segments target femur part against remaining femur part by the virtue of frame similarity for bottom up CT image sequences, so as to locate the position of the target femur accurately, and search for the edge point by the virtue of the segmentation carve. The result of the experiments show that the proposed method can extract the extern contour of proximal femur from CT images accurately with strong adaptivity and good anti-jamming.

Key words: hip joint, CT images, extern contour extraction

摘要： 为正确有效地分割出髋关节的外轮廓，提出了一种基于最佳分割线的股骨近端外轮廓的提取方法。该方法利用自底向上CT序列图像的帧间相似性找到一条分割目标股骨与其他骨骼的闭合分割线，从而利用该分割线来精确定位目标骨骼的位置，并搜索边缘点。实验结果表明，该方法能够准确地识别出股骨近端的外轮廓，自适应强，抗干扰性好。

关键词: 髋关节, CT图像, 外轮廓提取