Multi-kernel CNN block-based detection for COVID-19 with imbalance dataset

COVID-19, which originated from Wuhan, rapidly spread throughout the world and became a public health crisis. Recognizing positive cases at earliest stage was crucial in order to restrain of this virus perform medical treatment quickly for patients affected. However, limited supply RT-PCR as diagnosis tool caused greatly delay obtaining examination results suspected patients. Previous research stated that using radiologic images could be utilized detect COVID-19 before symptoms appeared. With rapid development Artificial intelligence imaging recent years, deep learning core technology achieve human-level-performance diagnostic accuracy. In paper, implemented chest X-ray dataset. The proposed model employed multi-kernel convolution neural network (CNN) block combined with pre-trained ResNet-34 overcome an imbalanced adopted different kernel sizes follows 1x1, 3x3, 5x5, 7x7. findings show is capable performing binary three class classification tasks accuracy 100% 93.51% validation phase 95% 83% test phase, respectively.