Evolving deep convolutional neutral network by hybrid sine-cosine and extreme learning machine for real-time COVID19 diagnosis from X-ray images

Abstract

The COVID19 pandemic globally and significantly has affected the life and health of many communities. The early detection of infected patients is effective in fighting COVID19. Using radiology (X-Ray) images is, perhaps, the fastest way to diagnose the patients. Thereby, deep Convolutional Neural Networks (CNNs) can be considered as applicable tools to diagnose COVID19 positive cases. Due to the complicated architecture of a deep CNN, its real-time training and testing become a challenging problem. This paper proposes using the Extreme Learning Machine (ELM) instead of the last fully connected layer to address this deficiency. However, the parameters' stochastic tuning of ELM's supervised section causes the final model unreliability. Therefore, to cope with this problem and maintain network reliability, the sine-cosine algorithm was utilized to tune the ELM's parameters. The designed network is then benchmarked on the COVID-Xray-5k dataset, and the results are verified by a comparative study with canonical deep CNN, ELM optimized by cuckoo search, ELM optimized by genetic algorithm, and ELM optimized by whale optimization algorithm. The proposed approach outperforms comparative benchmarks with a final accuracy of 98.83% on the COVID-Xray-5k dataset, leading to a relative error reduction of 2.33% compared to a canonical deep CNN. Even more critical, the designed network's training time is only 0.9421 ms and the overall detection test time for 3100 images is 2.721 s.

Keywords: COVID19; Chest X-ray images; Deep convolutional neural networks; Extreme learning machine; Sine–cosine algorithm.

Fig. 1

The architecture of LeNet-5 deep CNN

Fig. 2

A single-hidden layer neural network

Fig. 3

The effect of sine and cosine functions on Eqs. (8) and (9)

Fig. 4

Changes in sinus and cosine functions in a specified interval

Fig. 5

Changes in the sinus and cosine functions within the range of [−2, 2] causes to get closer or more distant from the desired response

Fig. 6

Reduction in the range of sine and cosine functions during iterations

Fig. 7

Six stochastic sample images from the COVID-X-ray-5k dataset

Fig. 8

The conventional vs. proposed architecture

Fig. 9

Assigning the deep CNN’s parameters as the candid solution (searching agents) of SCA

Fig. 10

The pseudo-code for DCELM-SCA model

Fig. 11

The flowchart of the designed model

Fig. 12

The EPG produced by in_6c_2p_12c_2p structure

Fig. 13

The EPG produced by in_8c_2p_16c_2p structure

Fig. 14

The confusion matrix for in_6c_2p_12c_2p model

Fig. 15

The confusion matrix for in_8c_2p_16c_2p model

Fig. 16

The ROC curves and precision-recall curves for DCELM-SCA and other benchmarks

Fig. 17

Level trends of the analyzed parameters

Fig. 18

Search history, convergence curve, average fitness history, and trajectory of some functions

Fig. 18

Search history, convergence curve, average fitness history, and trajectory of some functions

Fig. 19

ROI for positive COVID19 cases using ACM

Fig. 20

ROI for Normal cases using ACM