Segmentation of microscopic images of sputum stained by Ziehl - Nielsen using wavelet transform Mexican Hat

Abstract

Background. Currently at bacterioscopic diagnosis of tuberculosis there is a large number of errors that is associated with a number of problems that can be solved by automated analysis of microscopic images of sputum. Automated image analysis usually consists of several stages: image segmentation and identification or recognition of objects in the image. The article examines the first of these stages - segmentation. The aim of the study was to investigate the possibility of segmenting a digital image of a microscopic preparation sputum, stained by the method of Ziehl - Nielsen, using wavelet transform Mexican Hat. Materials and methods. As research material we used 830 digital images obtained by microscopy of sputum smears stained by the method of Ziehl - Nielsen. For the automated segmentation of images we used two-dimensional wavelet transform of Mexican Hat Wavelet. Results. During the study we defined the optimal value of the a parameter, which is the only varying parameter of a wavelet Mexican Hat, and carried out the run-time evaluation of the wavelet transform digital microscopic images of sputum stained by the method of Ziehl - Nielsen. Conclusions. The conclusion is made about possibility of using two-dimensional wavelet transform Mexican Hat digital microscopic images of sputum stained by the method of Ziehl - Nielsen, for segmentation of these images.

Keywords

• medical imaging
• image analysis
• automatic data processing
• bacteriological technic
• pulmonary tuberculosis