Search

COVID-19, is a dangerous disease, that is widely spread among humans by inhalation of the virus, and it harms and may damage the lung. The aim of this paper is to detect COVID-19 using our new algorithm called "Cascade-Correlation Growing Deep Learning Neural Network Algorithm (CCGDLNN)" from Computed tomography (CT) scan images of a patient's chest. We apply the algorithm over 48,260 computed tomography scan images from 377 persons divided into 282 normal persons and 95 patients were infected by COVID-19. Our system is divided into two stages: Firstly, the system removes unclear computed tomography-scan lung images by analyzing them. Secondly, we run our algorithm based on the exception model that begins with a small network without any hidden layers but has input and output layers only. The algorithm after that, adds new neurons and connects them to the last layer or add a new layer with one neuron. Finally, after performing these two stages, the system can be able to detect COVID-19 patients from their lung computed tomography-scan images. We train the data using two different models and compared the results with our model. In the image classification process, our model achieved 98.8% accuracy on more than 7996 test images. [ABSTRACT FROM AUTHOR]

In this paper, a proposed algorithm that dynamically changes the neural network structure is presented. The structure is changed based on some features in the cascade correlation algorithm. Cascade correlation is an important algorithm that is used to solve the actual problem by artificial neural networks as a new architecture and supervised learning algorithm. This process optimizes the architectures of the network which intends to accelerate the learning process and produce better performance in generalization. Many researchers have to date proposed several growing algorithms to optimize the feedforward neural network architectures. The proposed algorithm has been tested on various medical data sets. The results prove that the proposed algorithm is a better method to evaluate the accuracy and flexibility resulting from it. [ABSTRACT FROM AUTHOR]