Universitat Politecnica de Valencia, Plaza Ferrandiz y Carbonell, n.2, Alcoy (Alicante), 03801, Spain.
Math Biosci Eng. 2013 Apr;10(2):279-94. doi: 10.3934/mbe.2013.10.279.
Accurate image segmentation is used in medical diagnosis since this technique is a noninvasive pre-processing step for biomedical treatment. In this work we present an efficient segmentation method for medical image analysis. In particular, with this method blood cells can be segmented. For that, we combine the wavelet transform with morphological operations. Moreover, the wavelet thresholding technique is used to eliminate the noise and prepare the image for suitable segmentation. In wavelet denoising we determine the best wavelet that shows a segmentation with the largest area in the cell. We study different wavelet families and we conclude that the wavelet db1 is the best and it can serve for posterior works on blood pathologies. The proposed method generates goods results when it is applied on several images. Finally, the proposed algorithm made in MatLab environment is verified for a selected blood cells.
准确的图像分割在医学诊断中被广泛应用，因为该技术是生物医学治疗的一种非侵入性预处理步骤。在这项工作中，我们提出了一种用于医学图像分析的高效分割方法。特别是，通过这种方法可以对血细胞进行分割。为此，我们将小波变换与形态学操作相结合。此外，还使用小波阈值技术来消除噪声并为合适的分割做准备。在小波去噪中，我们确定最佳的小波，以便在细胞中显示出具有最大面积的分割。我们研究了不同的小波族，得出结论，db1 小波是最佳的，它可以用于后续的血液病理学研究。该方法在应用于多个图像时会产生良好的效果。最后，在 MatLab 环境中验证了所提出的算法对选定的血细胞的有效性。
