A High-Performance CNN Hardware Accelerator for Covid-19 Feature Extraction from X-ray Images

Covid-19 is a rapidly spreading disease with the potential to pose a life-threatening risk, so making prompt and accurate diagnosis of Covid-19-infected patients critically important. One method for detecting and diagnosing Covid-19 is using Convolutional Neural Networks (CNNs) to extract features from chest X-ray images. However, CNNs involve high computational complexity, requiring billions of operations per image and prolonged processing times. The aim of this paper is to design a hardware accelerator for a CNN to efficiently extract features indicative of Covid-19 from chest X-ray images. We have designed and simulated convolution unit and max-pooling unit, thereby constructing a CNN hardware accelerator for Covid-19 feature extraction from chest X-ray images. Simulation results demonstrate that the convolution unit and max-pooling unit function according to the algorithm’s specifications. Validation results also indicate the effective operation of the CNN hardware accelerator, enabling rapid and accurate detection and classification of Covid-19. With the support of the hardware accelerator, computational performance is optimized, and energy consumption is reduced, thereby enhancing the diagnostic speed of the system.