Segmentation of nucleus and cytoplasm of white blood cells using Gram-Schmidt orthogonalization and deformable models

Automatic recognition of white blood cells in hematological can be divided into four major parts: preprocessing, image segmentation, feature extraction and classification. Due to the multifarious nature of these cells and uncertainty in the hematological images, segmentation of white blood cells is one of the most important stages in this process. A scrupulous segmentation obviously reduces errors of next stages. In this paper, we introduce a novel method based on Gram-Schmidt process and parametric deformable models for segmenting the nucleus and cytoplasm. Also, we propose a new preprocessing method for improving the results of cytoplasm segmentation. Moreover, for finding the initial contour for parametric deformable model, an automatic scheme is defined. Experimental results show that our proposed method is capable of segmenting the white blood cells in the hematological images. To evaluate the proposed algorithm quantitatively, we compare its results with the manual segmentations by a hematologist. This study shows robustness of the proposed method. Another feature of the proposed method is that it is simple to implement.