ABSTRACT
In the medical field, early diagnosis means the definition of the disease in the period when the clinical signs of the disease do not appear or there are symptoms that do not cause pain and distress to the person. Delayed diagnosis causes aggravation of the disease and even the disappearance of the possibility of treatment. The field of medical diagnostics contains a multitude of challenges that are very similar to classical machine learning problems. Particularly in the field of radiology, there are multi-label classification problems where medical images are interpreted to indicate multiple existing or suspected pathologies. Chest X-ray is preferred in the diagnosis of chest diseases because it is inexpensive, widely available, and uses low-dose radiation. However, in many countries, the insufficient number of radiologists and difficulties in making a diagnosis reduce the success rate. This shows the need for an artificial intelligence system that can diagnose from chest X-rays. In this study, a deep learning model has been developed that can diagnose from chest X-rays. Training and testing of the developed convolutional neural network model was carried out with the ChestX-ray14 dataset consisting of 112,120 chest X-rays taken from 30,805 different patients. As a result of the experimental studies, the developed deep learning model was compared with the literature and it was seen that it achieved successful results.
