Table 4.

Comparison results between B-DDLN and some existing state-of-the-art methods.

Model/method	Testing accuracy ($\%$)
The proposed B-DDLN (case 1)	98.8889
The proposed B-DDLN (case 2)	98.8889
ResNet	97.2222
CNN-X (transfer learning)16	95.0000
ResNet-based multi-channel transfer learning model50	93.8889
DarkCovidNet51	97.7778
SPEA-II-based modified AlexNet52	98.3333
Ensemble densely connected convolutional neural network53	97.2222
Ensemble deep transfer learning model54	96.1111

The superior performance of the proposed B-DDLN diagnosis model are highlighted in bold.