1. Early detection of diabetes mellitus using statistical texture feature in finger nail image.
- Author
-
Kurniastuti, Ima, Andini, Ary, and Nerisafitra, Paramitha
- Subjects
DIABETES ,FINGERNAILS ,DISTRIBUTION (Probability theory) ,FEATURE extraction ,PREDIABETIC state - Abstract
The aim of research was obtained range number of the statistical texture feature of finger nail image that can be feature in classification. This paper consist of three steps such as preprocessing, feature extraction, and group frequency distribution. Preprocessing consist of resize, cropping and grayscaling. Feature extraction is used was statistical texture feature that consist of mean, variance, skewness, kurtosis and entropy. For obtained range number of texture feature, this research used group frequency distribution. The result showed that the range number of mean for normal data is 119 – 126, for prediabetes data is 137-145 and for diabetes data is 127 – 135. In variance, its range number for normal data was about 3802 – 4486, for prediabetes data was about 3734 – 4614 and for diabetes data was about 3175 – 3892. In skewness, its range number for normal data was about 0,8 – 1,02, for prediabetes data was about 1,21 – 1,44 and for diabetes data was about 1,22 – 1,42. And for kurtosis, the range number in normal data around (-0,56) (-0,08), in prediabetes data around (0,44) – (-0,07) and in diabetes data around (-0,24) – 0,31. Whereas in entropy, range number for normal data is 5,46 -5,69, for prediabetes data around 5,39 – 6,07 and for diabetes data around 5,27 – 5,49. In the future works, combination of two features or more in finger nail image could be used as feature for classification. [ABSTRACT FROM AUTHOR]
- Published
- 2023
- Full Text
- View/download PDF