<![CDATA[Background: Pneumonia is a respiratory infection that can be life-threatening if not properly diagnosed and treated. The diagnosis of pneumonia currently relies on the expertise of pulmonologists to evaluate chest X-ray results. Therefore, there is a need for technology that can assist doctors in analyzing X-ray images quickly and accurately. Methods: This study employs Convolutional Neural Networks (CNN) to classify chest X-ray images into three classes: Normal, Mild Pneumonia, and Severe Pneumonia. Several experiments were conducted by varying the number of epochs, dataset size, image resolution, and the number of hidden layers to achieve accurate identification results. Results: The final testing results showed that using 15 epochs, 5 hidden layers, and a dataset of 5700 images for classification with CNN can achieve a training accuracy of 92.48% and a validation accuracy of 91%. Results from 50 chest X-ray images indicated identical identification accuracy between the readings by doctors and the proposed method, with doctors taking 15 minutes to read and the proposed method taking only 0.2 seconds with an identification accuracy of 100%. Conclusion: This study demonstrates that the proposed method can assist pulmonologists in diagnosing pneumonia with high diagnostic accuracy and short diagnosis time, thereby helping to improve the quality of healthcare services. Recommendations: This study recommends the use of CNN as a method for diagnosing pneumonia in chest X-ray images.]]>
Copyrights © 2024
