Table 4 Test confusion matrices for the deep neural networks in COVID-19 detection (o.o.d. evaluation)

From: Improving deep neural network generalization and robustness to background bias via layer-wise relevance propagation optimization

True class

Predicted class

  
 

Normal

Pneumonia

COVID-19

ISNet

 Normal

210

157

3

 Pneumonia

81

1210

4

 COVID-19

93

156

1266

U-Net + DenseNet121

 Normal

296

9

65

 Pneumonia

271

604

420

 COVID-19

95

149

1271

DenseNet121

 Normal

211

16

143

 Pneumonia

306

381

608

 COVID-19

63

62

1390

RRR

 Normal

213

84

73

 Pneumonia

355

576

364

 COVID-19

243

130

1142

Multi-task U-Net

 Normal

125

42

203

 Pneumonia

62

103

1130

 COVID-19

38

300

1177

Attention Gated Sononet

 Normal

57

2

311

 Pneumonia

346

233

716

 COVID-19

151

114

1250

Extended GAIN

 Normal

81

145

144

 Pneumonia

241

526

528

 COVID-19

108

200

1207

Vision Transformer

 Normal

247

50

73

 Pneumonia

279

538

478

 COVID-19

393

385

737

Back to article page