Recognition and Classification of Childhood Pneumonia Based on Improved Inception-ResNet-v2

Table 1. Number of grades of pneumonia in children
View table
Table 1. Number of grades of pneumonia in children
Category Type of pneumonia Quantity
0 Bacterial pneumonia 2780
1 Normal 1583
2 Viral pneumonia 1493

Table 2. Classification confusion matrix of fusion iteration attention feature fusion module and improved Inception-ResNet-v2
View table
Table 2. Classification confusion matrix of fusion iteration attention feature fusion module and improved Inception-ResNet-v2
Category 0 1 2
0 260 5 43
1 3 151 2
2 28 2 92

Table 3. Comparison of two classification experiments
View table
Table 3. Comparison of two classification experiments
Method $R_{a c c u r a c y}$ /%
Teacher module 94.87
Method of reference［10］ 96.39
GIV3 96.77
Proposed method 97.9

Table 4. Comparison of experimental results

View table

Table 4. Comparison of experimental results

Method	Category	R_accuracy /%	R_precision /%	R_recall /%	$R_{s p e c i f i c i t y}$ /%
VGG16	bacteria	79.5	82.1	76.4	85.8
	normal		94.2	93.8	97.7
	virus		61.0	68.0	83.8
ResNet50	bacteria	81.3	79.8	80.2	81.9
	normal		93.3	94.6	97.3
	virus		66.3	63.4	88.7
Inception-v3	bacteria	82.2	80.7	86.0	82.7
	normal		95.1	95.7	98.1
	virus		66.6	62.8	91.0
Inception-ResNet-v1	bacteria	78.8	76.7	86.0	76.2
	normal		94.3	93.6	98.2
	virus		67.7	54.2	89.8
Inception-ResNet-v2	bacteria	79.1	77.4	86.0	78.4
	normal		91.9	74.2	97.0
	virus		67.2	52.8	90.4
Proposed method	bacteria	85.8	84.4	89.3	83.7
	normal		96.8	95.6	98.8
	virus		75.4	67.2%	93.3%

Table 5. Comparison of ablation experiment results

View table

Table 5. Comparison of ablation experiment results

Number	Method	R_accuracy /%	R_precision /%	R_recall /%	$R_{s p e c i f i c i t y}$ /%
Experiment 1	Inception-ResNet-v2	79.1	78.8	78.4	88.6
Experiment 2	Inception-ResNet-v2+SiLU	82.3	81.3	82	90.4
Experiment 3	Inception-ResNet-v2+new stem	83.4	83.6	83.3	90.7
Experiment 4	Inception-ResNet-v2+MS_CAM	83.0	82.2	82.2	90.8
Experiment 5	Proposed method	85.8	85.5	84.0	95.2

Table 6. Comparison of experimental results of the COVID-19 dataset
View table
Table 6. Comparison of experimental results of the COVID-19 dataset
Method R_accuracy /%
Method of reference［31］ 93.0
Method of reference［30］ 97.0
Proposed method 98.9

Tools

Get Citation

Copy Citation Text

Junhao Yang, Zhiqing Ma, Guohui Wei, Shuang Zhao. Recognition and Classification of Childhood Pneumonia Based on Improved Inception-ResNet-v2[J]. Laser & Optoelectronics Progress, 2023, 60(14): 1410008

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites