COVID-19 Patients Detection in Chest X-ray Images via MCFF-Net

COVID-19 is a respiratory disease caused by severe acute respiratory syndrome coronavirus (SARS-CoV-2). This paper proposes a deep learning model to assist medical imaging physicians in diagnosing COVID-19 cases. We designed the Parallel Channel Attention Feature Fusion Module (PCAF), and brand new structure of convolutional neural network MCFF-Net was put forward. The experimental results show that the overall accuracy of MCFF-Net66-Conv1-GAP model is 96.79% for 3-class classification. Simultaneously, the precision, recall, specificity and the sensitivity for COVID-19 are both 100%. Compared with the latest state-of-art methods, the experimental results of our proposed method indicate its uniqueness.