Identification of Bacterial Pneumonia on Chest X-Ray Images Using InceptionV3 Feature Extraction and K-Nearest Neighbor (K-NN)

Bacterial pneumonia is a type of respiratory tract infection that can lead to serious complications if not promptly detected and properly treated. This study develops a bacterial pneumonia identification model using chest X-ray images by combining the InceptionV3 feature extraction method with the K-Nearest Neighbor (K-NN) classification algorithm. The process begins with image preprocessing to enhance visual quality, followed by feature extraction using InceptionV3 to capture texture and shape characteristics of the lung area. The extracted features are then classified using the K-NN algorithm. Based on the experiments, the highest classification accuracy was obtained at K = 3, reaching 0.84. Model consistency was further evaluated using a cross-validation scheme with odd K values ranging from 1 to 20, and the best result was achieved at K = 3 with an accuracy of 0.9455. The experimental results indicate that the combination of InceptionV3 and K-NN is effective and promising as an automatic diagnostic tool for detecting bacterial pneumonia through chest X-ray images.