基于标记控制分水岭算法的乳腺X线摄片分割

乳房区域提取是基于乳腺X线摄片的计算机辅助诊断中的关键步骤。它能够将病灶的检测范围限定在乳房区域之内,减少背景区域的干扰,从而提高诊断效率。针对乳房区域边缘处的灰度与背景区域很接近,难以区分的问题,提出一种基于标记控制分水岭算法的乳房区域分割方法。首先对图像进行形态学平滑并计算梯度图;然后,基于改进的Otsu阈值法及形态学方法确定内部标记和外部标记;最后,在内外标记的控制下对图像进行分水岭分割。采用多种评估方式将算法的分割结果与金标准进行对比,其重叠率达到0.93±0.03,结果表明该算法能有效提高乳房区域分割精度。

Mammogram Segmentation Based-on Marker-controlled Watershed Transform

Extraction of breast contour is a crucial step in computer aided diagnosis of mammograms. It has the advantage of allowing the search for abnormalities to be limited to the region of the breast without undue influence from the background. Breast region segmentation is difficult because of the tapering nature of the breast. A marker-controlled watershed method has been proposed for breast region segmentation on mammograms in this paper. First, the segmentation function is determined on the smoothed mammogram by morphological methods. Then, the internal and external markers are calculated on basis of a improved Otsu method and morphological operations. Results are evaluated by comparison with gold standard for a set of 120 mammograms. The mean ?std of the values of the area overlap metric for our method is 0.93 ?0.03. The experimental results indicate that the algorithm can effectively improve the accuracy of breast region segmentation on mammogram.