Determining The Threshold Value for Identification of The Goblet Cells in Chicken Small Intestine
Dedi Sepriana (a*), Kusworo Adi (b), Catur Edi Widodo (b)
a) Master of Information System, Diponegoro University
Semarang, Indonesia
*dedisepriana[at]gmail.com
b) Department of Physics, Diponegoro University
Semarang, Indonesia
Abstract
The health condition of chicken intestines can be seen from the number of goblet cells found in the small intestinal epithelium. The cell identification is required to count the number of goblet cells and in the most cases, identification of goblet cells is carried out through a microscope directly. Although the directly identification saves costs, but the implementation not only takes a lot of time and energy but also is relatively subjective. Here we seek to design a system that can identify goblet cells automatically through segmentation by using two threshold values. The two threshold values are obtained from the gray scale images. Gray scale image is defined as the result of extraction of the Hue component in HSV type images. Goblet cells as the main producer of mucin can be colored in the processing of tissue preparation by using AB-PAS staining technique. This technique turns out the mucin into blue, so the identification of goblet cells can automatically be conducted by extracting blue features. After that we decide the lower and upper threshold values from the results of extraction which show the object as a goblet cell. In conclusion, we provide the results that the threshold value to separating object from other networks is on the uint8 gray scale, with value 176 to 221. The segmentation using two threshold values proves that the system is able to identify the goblet cells properly.
Keywords: Goblet cell; RGB to HSV conversion; Hue component extraction; Multilevel thresholding
Topic: International Symposium of Engineering, Technology, and Health Sciences