基于钙化点信息的乳腺病灶类型识别算法

Recognition Algorithm of Breast Disease Focus Based on Calcifications Information

为了有效地对乳腺恶性病变做出早期诊断,提出一种基于钙化点信息的乳腺病灶类型识别算法。算法以钙化点的形状信息、纹理信息、空间分布信息为特征量,以支持向量机为工具对钙化点进行恶性与良性分类。对开放乳腺图像数据库MIAS的仿真实验表明,算法的检出率超过86%,错检率不足4%,达到理想的识别效果。

In this paper,a new recognition algorithm of breast disease focus based on calcifications information is proposed in order to effectively diagnose the early malignant breast disease.In this recognition algorithm,the characteristic quantities are shape information,texture information and space distribution information of the calcifications and the tool supported vector machine is employed to classify the calcifications is benign or malignant.The image samples from the mammogram database MIAS are detected by this algorithm.The simulation results are that the rate of detection calcifications was more than 86% and the false positive was less than 4%,which show the effectiveness of the proposed recognition algorithm.