Xception Model for Pneumothorax Classification using Chest X-ray Images
Rahul Singh, Avinash Sharma, Neha Sharma, Rupesh Gupta
Abstract
Pneumothorax is a disease that might threaten the patient’s life and must be recognized and treated as soon as feasible. An X-ray of the chest should be conducted as the first diagnostic test when a pneumothorax is suspected. Due to the advent of deep learning techniques, its diagnosis can be accurately performed. If it can be used successfully, it may be able to help us find solutions to a range of problems. The application of classification in the prognosis of pneumothorax greatly aids medical professionals in treating the illness. It took a long time to uncover the problem, which caused numerous pneumothorax patients’ deaths. This unique technology enables us to diagnose patients more swiftly and accurately. A pre-trained model called Xception was used in this work. The model performs admirably, and the time spent training decreases thanks to the pretrained models. The model is performed on a total of 2027 photos, with 1824 images used for training, 102 for testing, and 101 for validation. Precision, recall, accuracy, and F1-score are used to calculate the findings. When the model’s parameters are compared, the total accuracy is determined to be 85%.