基于Tirm序列联合直方图及纹理特征分析鉴别乳腺良恶性病变
DOI: 10.3971/j.issn.1000-8578.2018.18.0815
Keywords: A Case Report and Literature Review,Primary Sebaceous Carcinoma of Mandibular Gingiva: A Case Report and Literature Review,An Modified Primary Culture Method of Breast Cancer-associated Fibroblasts,Clinical and MRI Features of Tumors of Septum Pellucidum,Value of 18F-FDG Distribution for Differential Diagnosis of High Metabolic Lung Lesions,Efficacy and Safety of Oral Etoposide on Heavily Pretreated Metastatic Breast Cancer,Imaging Features of Primary Hemangiopericytoma of Bone,NDRG2 Inhibits Migration of Breast Cancer Cells Through Upregulating PTEN Expression,Metaplastic Carcinoma of Breast with Mesenchymal Differentiation: 28 Cases Report,MDM2 siRNA Inhibits Angiogenesis by Reducing Vascular Endothelial Growth Factor Production in Breast Cancer Cells,Complementarity Determining Region 3 Repertoire of Clonal T Cell Receptor α Chain in Patients with Micrometastasis from Breast Cancer,Difference of Self-renewal and Tumorigenicity among Different Breast Cancer Stem Cells Subpopulations,Relationship of ALDH1+/CD133+ Stem Cell-like Cells with Angiogenesis in Breast Invasive Ductal Carcinoma Tissues,Accuracy of Mammography, Color Doppler Ultrasonography and Magnetic Resonance Imaging in Predicting Ductal Carcinoma in Situ
Abstract:
摘要 目的 分析乳腺良恶性病变在T2反转恢复压脂序列上的直方图及纹理特征，并评价诊断乳腺良恶性病变的效能。方法 回顾性分析100例行MRI检查并经病理证实的乳腺肿瘤资料。纹理特征提取采用灰度共生矩阵法得到病灶能量值、熵值、惯量值、自相关性、逆差矩及群显著性；直方图分析得到平均值、偏度、峰度。分别评价不同参数值在乳腺良恶性病变中的诊断效能。构建受试者工作特征曲线（ROC）结合临床实际评价乳腺良恶性组间的差异，确定诊断临界值。结果 各参数在良恶性组间比较，自相关性、群显著性、平均值及峰度值差异均有统计学意义（均P<0.001）；能量值、熵值、惯量值、逆差距及偏度值差异无统计学意义（P>0.05）。构建ROC将自相关性、群显著性、平均值及峰度值分别用于诊断乳腺良恶性病变，均有统计学意义（P<0.001），联合四组参数诊断良恶性病变，曲线下面积为0.868，敏感度为90.57%，特异性为72.34%。结论 基于Tirm序列的直方图及纹理分析可用于乳腺良恶性病变的鉴别诊断，自相关性、群显著性、平均值及峰度值在Tirm序列上对诊断乳腺良恶性病变有一定意义，联合诊断可提高诊断效能
Full-Text
