Pneumonia detection and classification using lung X-ray image features with machine learning algorithms

Pneumonia is a disease that threatens human health and can lead to death when diagnosed late. In this study, pneumonia detection was performed using machine learning methods from lung X-ray images. The dataset consists of lung X-ray images of children aged 1-5 years. A new approach was presented for pneumonia detection. In the first step, a series of preprocessing steps were performed on the lung X-ray images. These steps are Histogram Equalization, Mask R-CNN(Mask Region-Based Convolutional Neural Network), and Otsu Thresholding. Then, first-order and second-order textural features were extracted. By using the SMOTE(Synthetic Minority Over-sampling Technique) method, the imbalance between the classes in the training data was addressed. In the last step, classification was performed using machine learning classification models. These classification models are Multilayer Perceptron, Support Vector Machine, Logistic Regression and K-Nearest Neighbors. Classification models were compared with each other. According to the results, the Multilayer Perceptron showed the best performance compared to other classification models, with an accuracy rate of 0,95833, a precision rate of 0,95978, a recall rate of 0,95833, an F1-Score of 0,95856 and an AUC(Area Under Curve) Score of 0,98566. In addition, classification models were compared with previous studies in the literature. According to the results, the proposed Multilayer Perceptron outperformed the other studies in the literature in terms of accuracy, F1-Score and AUC Score.