Open AccessAutomated Intelligent Diagnostic Procedure for Chronic Leukemia
Author(s) -
Chan Lok Pooi,
Mohd Yusoff Mashor,
R. Adollah,
Rosline Hassan
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1372/1/012007
Subject(s) - thresholding , artificial intelligence , pattern recognition (psychology) , feature selection , computer science , classifier (uml) , feature extraction , segmentation , chronic leukemia , image segmentation , perceptron , artificial neural network , image (mathematics) , leukemia , acute leukemia , medicine
Due to growing statistics and the important role of early diagnosis for chronic Leukemia, an automated intelligent diagnostic procedure for chronic Leukemia is needed. This paper presents an automated procedure for this diagnostic system which consisted of four main stages; namely Image Segmentation, Feature Extraction, Feature Selection and Classification. Colour Thresholding and Gradient Edge Detection were applied to segment the nucleus of white blood cells from the image. A total of 548 cells nuclei were extracted from 100 images and used to develop and validate the proposed procedure. Three feature selection algorithms and three classifiers were analyzed and evaluated in order to obtain a reliable and robust automated diagnostic procedure. Based on the results, the ReliefF algorithm was selected and applied for feature selection, the method yields the best accuracy of 98.1% for testing accuracy. Besides, the selected classifier was Multilayer Perceptron (MLP) network trained by Levenberg Marquardt (LM) algorithm.