Skip to main content
. 2022 Jul 18;126:109319. doi: 10.1016/j.asoc.2022.109319

Table 6.

Summary of related studies for chest X-ray classification with pneumonia conditions.

Study DL technique Acc % Loss function Optimizer GPU Evaluation metric
[40] CNN 83.38 Cross-entropy Adam Acc
[22] CNN 96.18 CCE Adam Sn, Sp, P, F1-score,
Acc, AUROC
[23] DensetNet121
ResNet-50
InceptionV3
76
69
61
BCE Adam AUROC
[24] Ensemble model (ResNet18, DenseNet121, InceptionV3, Xception, MobileNetV2) 98.43 Cross-entropy SGD Acc, R, P, F1-Score, AUROC
[73] ResNet-50 97.65 Categorical Accuracy
[9] Ensemble model (AlexNet, DenseNet121, InceptionV3, ResNet18, GoogLeNet) 96.4 Cross-entropy Adam AUROC, R, P, Sp, Acc
[80] MobileNetV2 90 Cross-entropy Acc
[81] MobileNetV2 93.4 BCE Adam AUROC, Acc, Sp, Sn
[85] DenseNet121 76.80 Weighted BCE Adam F1-score
[55] Ensemble model (ResNet-34 based U-Net, EfficientNet-B4 based U-Net) 90 BCE, Dice loss Ranger optimizer Acc, P, R, F1-score
[101] CNN 93.73 Acc
[125] Mask RCNN
(ResNet-50+ResNet101)
Multi-task loss SGD IoU for true positive
[119] CNN 86 BCE Acc
[113] CNN 90.68 BCE Adam Acc
[114] CNN 97.34 Cross-entropy MSE Gradient Descent Acc
[68] VGG-16 with MLP 97.4 RMSprop Acc, Sn, Sp, AUROC, F1-score.
[126] CNN 98.46 P, R, Acc, F1-Score, AUROC, cross validation
[115] CNN 92.31 CCE Adam Acc, R, F1-score
[102] CNN 95.30 CCE Adam cross validation, Acc, AUROC
[128] CNN
MLP
94.40
92.16
Cross-entropy 5-fold cross validation, Acc, AUROC
[127] SCN 80.03 BCE Adam Acc, P, R, F1-score
[69] VGG-16
Xception
87
82
CCE RMSprop Acc, Sp, R, P, F1-score.
[86] DenseNet-169
with SVM
80.02 AUROC
[51] InceptionV3 with
more layers
90.1 CCE Nadam Acc, P, R, F1-score
[62] CNN with U-Net 97.8 Adam AUROC, Acc