超声图像定量分析对乳腺良、恶性肿瘤诊断价值.docx

【摘要】? 目的 研究乳腺良、恶性肿瘤超声诊断量化方法。方法 对手术确诊的良、恶性乳腺肿瘤各10例的原始超声图像的灰度特征进行提取，利用Matlab 7.0图像处理函数及算法编程 HYPERLINK /pc/ 计算两组图像的4个灰度特征参数(灰度均值M、灰度标准差V、扭曲度S、熵H)，对比分析各参数在2组样本之间的差异。结果 良、恶性乳腺肿瘤患者超声图像的灰度均值M、灰度标准差V、扭曲度S均存在显著性差异，分别为60.30±11.77 vs 82.64±5.35、27.49±4.67 vs 31.89±3.38、0.20±0.05 vs 0.85±0.22。结论 超声图像定量分析能量化乳腺良、恶性肿瘤超声图像的特征，M、V、S具有作为乳腺良、恶性肿瘤超声诊断量化指标的价值。 【关键词】? 乳腺肿瘤;超声诊断;图像定量分析;灰度特征 　　Abstract： Objective To investigate a feasible approach to the quantitative analysis of ultrasonic images in diagnosing benign and malignant breast tumors. Methods The original ultrasonic images of 10 cases of benign breast tumors and 10 cases of malignant breast tumors diagnosed by surgical operation were collected and the grey-scale features were extracted. The Matlab 7.0 image processing function and algorithm programming was used to calculate the four grey-scale feature parameters, e.g., mean (M), standard variance (V), skewness (S) and entropy (H) of the two groups of ultrasonic images. The differences in parameters between the two groups were analyzed. Results The differences of M (60.30±11.77 vs 82.64±5.35), V (27.49±4.67 vs 31.89±3.38) and S (0.20±0.05 vs 0.859±0.22) between the two groups were significant. Conclusion Quantitative analysis of ultrasonic images helps to quantize the grey-scale feature of ultrasonic images between benign and malignant breast tumors. The three parameters (M, V, S) are valuable quantitative indexes to ultrasonic diagnosis between benign and malignant breast tumors. 　　Key words：breast tumor; ultrasonic diagnosis; quantitative analysis of image; grey-scale feature 　　乳腺癌发病率、病死率均居女性恶性肿瘤首位，早期诊断对预后尤为重要。超声检查以其无创优点成为门诊疑似患者常用诊断方法，超声检查根据乳腺肿瘤影像的边界、形态和回声分布等特征判别其良恶性;但对于声像图特征不明显的乳腺肿瘤，诊断主要依赖医生的经验，而没有量化指标，因此诊断的准确性不能令人满意。本研究利用Matlab 7.0图像处理函数及算法编程，对乳腺良、恶性肿瘤超声图像的灰度特征参数进行对比分析，以期为乳腺肿瘤的良、恶性判别提供量化诊断依据。 　　1 资料和方法 　　1.1 图像采集 采集2008年1月—2009年1月徐州医学院附属 HYPERLINK /Medicine/ 医院收治经病理确诊的乳腺良、恶性肿瘤各10例患者的原始超声图像，经超声工作站保存在计算机中。超声诊断仪器为麦迪逊ACCUVIX XQ型，L5-12 MHz变频探头。 　　1.2 感兴趣区域选取 首先截取肿瘤组织部分的感兴趣