Detection Technique of Squamous Epithelial Cells in Sputum Slide Images using Image Processing Analysis

Fatin Izzwani Azman (Universiti Malaysia Pahang, Faculty of Electrical and Electronics Engineering, Pahang Darul Makmur)
Kamarul Hawari Ghazali (Universiti Malaysia Pahang, Faculty of Electrical and Electronics Engineering, Pahang Darul Makmur)
Rosyati Hamid (Universiti Malaysia Pahang, Faculty of Electrical and Electronics Engineering, Pahang Darul Makmur)
Noor Amira Syuhada Mahamad Salleh (Universiti Malaysia Pahang, Faculty of Electrical and Electronics Engineering, Pahang Darul Makmur)

Proceeding of the Electrical Engineering Computer Science and Informatics
Vol 1: EECSI 2014

A good quality sputum is important to detect diseases. The presence of squamous epithelial cells (SEC) in sputum slide images is important to determine the quality of sputum. The presence of overlapping SEC in sputum slide images causes the process become complicated and tedious. Therefore this paper discusses on technique of detection and summation for Squamous Epithelial Cell (SEC) in sputum slide image. We addressed the detection problem by combining K-means and color thresholding algorithm. The design of aided system is evaluated using 200 images and the proposed technique is capable to detect and count each SEC from overlapping SEC image. Total of 200 images were clustered to 10 groups, labelled as Group Cell 1 to group Cell 10 that correspond to the number of cells in the image. Therefore, each group will contain 20 images. The accuracy of the algorithm to detect SEC was also measured, and results show that in 91% which provides a correct SEC detection and summation.