基于灰度阈值分割和动态规划的超声图像乳腺肿瘤边缘提取

目的 寻求提高超声图像乳腺肿瘤边缘提取效果的最优方法。方法 先根据超声图像的灰度分布采用灰度闽值分割定位法提取乳腺肿瘤的初始边缘，然后根据图像灰度的梯度信息采用动态规划法进行边缘的修正，从而准确提取超声图像乳腺肿瘤的边缘。结果 对18例超声图像进行乳腺肿瘤的边缘提取，结果显示本文方法相比单一的灰度阈值分割法或单一的动态规划法能更为准确地提取超声图像乳腺肿瘤的边缘。结论 本文方法可以有效地用于超声图像乳腺肿瘤的边缘提取。

Boundary Extraction of Ultrasonic Breast Tumor Image Based on Gray-level Threshold Segmentation and Dynamic Programming

Objective To improve the accuracy of the boundary extraction method for ultrasonic breast tumor image identification. Method An initial boundary of the tumor was obtained by using gray-level threshold segmentation based on ultrasonic image intensity. To achieve a more accurate result, a dynamic programming method according to the image gradient was applied to adjust the initial boundary. Result Experiments on 18 ultrasonic images showed that this proposed method could extract the breast tumor boundary more accurately than either gray-level threshold segmentation or dynamic programming method. Conclusion The method presented in this paper can be used to extract the breast tumor boundary from ultrasonic image effectively.