基于灰度共生矩阵的骨肉瘤图像分析

Osseous tumors imaging analysis based on the gray level co-occurrence matrix

目的寻求骨肉瘤影像诊断的数字化特征指标。方法通过探讨利用基于灰度共生矩阵的图像纹理分析方法,对骨肉瘤影像的病变骨质区、正常骨质区、软组织区分别提取对比度、逆差矩、能量、熵和相关性5种特征值,进行数字特征分析。结果经过统计分析,病变骨区域、正常骨区域以及软组织区域三者之间的每一种特征值差异均有统计学意义。结论数字特征的提取对识别病变和正常组织有意义。

Objective To seek the digital characteristics for osseous tumor image diagnosis. Methods The contrast, homogeneity, energy, entropy and correlation were fetched from pathologically changed osseous region, normal osseous region and parenchyma region of the X-ray images with imge texture analysis methods based on the grey level co-occurrence matrix in 15 patients with osseous tumor. Results Significant differences of characteristics were detected among pathologically changed osseous area, normal osseous area and parenchyma area of the osseous tumor. Conclusion Digital characteristics fetched from image are significant for recognizing pathologically changed osseous and normal osseous.