Integrated image processing analysis and Naïve Bayes Classifier method for lungs X-ray image classification

To diagnose the lungs condition, commonly, the radiologists analyze the lungs image purely based on the X-ray image result with the naked eye. Hence, this method leads the subjectivity issue. In this study, the combination of image processing analysis and Naïve Bayes Classifier (NBC) are expected to overcome the issue. In the image processing analysis, we used the median filter and adaptive histogram equalization to enhance the lungs X-ray image quality. The five image features i.e., the feature mean, the feature SD, the feature kurtosis, the feature skewness, and the feature entropy were determined to obtain the characteristic of each lungs condition i.e., the normal lungs, the pleural effusion, and the lung cancer. In the NBC method, the five image features were used as the predictors to determine the lungs class i.e., the normal lungs class, the pleural effusion class, and the lung cancer class. The classification using NBC method consisted of two processes i.e., the training process and the validation process. The training process included the total numbers of 90 lungs X-ray images, whereas the validation process used the total numbers of 60 lungs X-ray images. According to the numerical calculation in the validation process, the performance of NBC method has 70% accuracy.