Kruggel Frithjof, Paul Joseph Suresh, Gertz Hermann-Josef
204 Rockwell Engineering Center, University of California, Irvine, CA 92697-2755, USA.
Neuroimage. 2008 Feb 1;39(3):987-96. doi: 10.1016/j.neuroimage.2007.09.058. Epub 2007 Oct 11.
Diffuse lesions of the white matter of the human brain are common pathological findings in magnetic resonance images of elderly subjects. These lesions are typically caused by small vessel diseases (e.g., due to hypertension, diabetes), and related to cognitive decline. Because these lesions are inhomogeneous, unsharp, and faint, but show an intensity pattern that is different from the adjacent healthy tissue, a segmentation based on texture properties is proposed here. This method was successfully applied to a set of 116 image data sets of elderly subjects. Quantitative measures for the lesion load are derived that compare well with results from experts that visually rated lesions on a semiquantitative scale. Texture-based segmentation can be considered as a general method for lesion segmentation, and an outline for adapting this method to similar problems is presented.
人脑白质的弥漫性病变是老年受试者磁共振图像中常见的病理表现。这些病变通常由小血管疾病（如高血压、糖尿病所致）引起，且与认知衰退有关。由于这些病变不均匀、边界模糊且较淡，但呈现出与相邻健康组织不同的强度模式，因此本文提出了一种基于纹理特征的分割方法。该方法已成功应用于116例老年受试者的图像数据集。得出了病变负荷的定量测量结果，与专家在半定量尺度上对病变进行视觉评分的结果具有良好的可比性。基于纹理的分割可被视为病变分割的通用方法，并给出了将该方法应用于类似问题的概述。
