HEP-2 CELL FEATURE EXTRACTION USING WAVELET AND INDEPENDENT COMPONENT ANALYSIS

Human antibodies work to attack any diseases or bacteria that presented inside the body. However, there is an act when human antibodies tend to attack own body cells or tissues which is called as Anti-nuclear Antibodies (ANA). ANA consist of many different types that can be recognized by its nucleus size and shape. Common method of classifying ANA is by performing Indirect Immunofluorescences (IIF) with HEp-2 cell and observed the pattern under the microscope by naked eye which said to be inaccurate, takes time and subjective. Thus, this project will study on the technique to identify and classify the pattern of ANA automatically. Algorithms are created using MATLAB software and a Graphical User Interface (GUI) is generated for the algorithm to be easily used. This work will focus more on feature extraction using Wavelet and Independent Component Analysis (ICA). The type of Wavelet Transform that will be used is the 2D Discrete Wavelet Transform (2D DWT) and Fast ICA for Independent Component Analysis. Then Support Vector Machine (SVM) is used to perform the classifications parts using the features extracted from both methods. Different features obtained are tested in SVM and the performance of both methods is compared. From the result, it shows that by using the same classifier, Wavelet can provide better features for classification compared to ICA.