目的 利用灰度-基元共生矩阵法研究多发性硬化(MS)MRI图像脑白质中肉眼可见病灶区(L-W)、表现正常区(A-W)及健康对照组脑白质区(N-W)的纹理特征,目的在于测试A-W与N-W的纹理特征间是否存在显著性差异。方法 从26例MS患者MR的T2WI图像中提取L-W和A-W,从26例健康对照者提取与A-W对应部位的N-W,组成三组数据样本集。经灰度-基元共生矩阵映射,提取能量、对比度、逆差矩和熵等纹理参数,并比较三组纹理参数是否显著不同。结果 L-W、A-W、N-W三组样本的所有纹理参量均显示,A-W组与N-W组之间存在显著性差异,与L-W组也显著不同(P〈0.01)。结论 常规MR的T2WI图像纹理分析显示,MS患者侧脑室表现正常脑白质与健康对照组显著不同。通过纹理分析有可能揭示MS患者隐匿性病。
Objective To investigate the differences of texture characters among macroscopic lesion white matter(L-W), normal appearing white matter(A-W) in magnetic resonance images (MRI) from patients with multiple sclerosis(MS) and normal white matter(N-W) in healthy controls by gray-primitive co-occurrence matrix, and to detect the hidden abnormality of A-W. Methods L-W and A-W from twenty-six patients with MS and N-W form another twenty-six healthy controls regions matched A-W were selected, which were T2 weighted MR images. Regions of interest consist of L-W, A-W and N-W. All of the images were mapped into gray-primitive co-occurrence matrix and then texture characteristics of energy, contrast and entropy were calculated, and finally significance was tested among the three groups. Results In L-W, A-W and N-W groups, the increase tendency was noticed for energy (1.3033, 3.0201 and 4.3119) and contrast (2.1011, 73.0176 and 150.3742), while decrease tendency for inverse difference moment (2.1706, 0.1397 and 0.0646) and entropy (4.0033, 1.3515 and 0.5931).All differences for the four texture characters above were significant not only between A-W and N-W, but also between A-W and L-W (P〈0.01). Conclusion Texture characteristic analysis of T2 weighted MR images show the significant differences between A-W and N-W. Texture analysis can possible reveal the microscopic disease in patients with MS.