模式识别在乳腺超声弹性成像诊断中的应用研究

Application of Pattern Recognition in Breast Ultrasound Elastography

研究一种针对乳腺弹性超声影像的模式识别算法,并研究相关算法模型的临床价值。以支持向量机(SMO-SVM)为核心算法,结合基于SRAD模型和ICOV模型算法对弹性超声检查影像进行分析,使用基于水平集的算法做出初步判断,进而在GLCM移动窗口算法支持下,使用SMO-SVM神经网络对各移动窗口进行整合判断,对4种常见乳腺癌CA化灶进行判断。通过实证分析,该算法在对II期乳腺癌的判断中具有显著的临床价值,对I期乳腺癌的判断中也有一定的参考意义。

This paper studies a pattern recognition algorithm for breast elastography,and studies the clinical value of the algorithm model.With support vector machine(smo-svm)as the core algorithm,combined with SRAD model and icov model algorithm,the elastosonography images were analyzed,and the algorithm based on level set was used to make preliminary judgment.Then,with the support of GLCM moving window algorithm,smo-svm neural network was used to integrate the moving windows and judge the four common breast cancer CA lesions.Through empirical analysis,the algorithm has significant clinical value in the judgment of stage II breast cancer,and also has certain reference significance in the judgment of stage I breast cancer.