Optimization of characteristics using Artificial Neural Network for Classification of Type of Lung Cancer

Abstract

Early detection of lung cancer is a challenging problem because it is associated with the unique structure of cancer cells. In West Sumatra, Radiology Semen Padang Hospital and M Djamil general hospital of Padang as the hospital with the most complete facilities, the number of radiologists is only 4 radiologists. Even though they operate 24 hours, it condition is not enough in handling cancer detection. Accumulated CT scan results makes radiologists work not optimal. Human factors (human error) such as fatigue, not focusing on making the diagnosis wrong. For this reason, a system is needed that can assist radiologists in diagnosing CT scans that can help the radiologist to diagnose faster and reduce errors caused by human error. This paper presents the system using artificial neural network backpropagation method. This study resulted the artificial neural network backpropagation classification system to diagnose CT scans of lung cancer patients. This system has several steps and methods as part of the Computer Aided Diagnosis (CAD) system including segmentation of cancer images for simplifying data input, then feature extraction is done by processing data from the pixel value of the segmentation results by taking five characters, namely the number of areas, mean, standard deviation, curtosis and skewness as features or characteristics of the data. Then using the ANN backpropagation algorithm for the classification stage of cancer types. System testing shows that the results of the system accuracy with hospital diagnosis have training data accuracy of 88.89% and test data of 83.33%.