Cloud-Based Lung Tumor Detection and Stage Classification Using Deep Learning Techniques

Table 8

Comparison of performance metrics with existing methods.

Methods	Medical database	Classifier	%Accuracy	%Sensitivity	%Specificity	False positive rate (FPR)	AUC
Proposed method	LIDC-IDRI/PET	M-CNN	97.1	95.9	93.9	2.1	94
Eali Stephen et al. (2021)	LUNA	3D CNN-AlexNet detection algorithm	89	83.5	80.5	—	91
Tafadzwa et al. (2021)	LUAD	Supervised CNN predictor	90	81.5	78.5	—	71%
Pragya et al. [17]	LIDC-IDRI, LUNA 16	SVM, KNN, and CNN	91	87.2	84.2	—	—
Kalaivani et al. (2021)	LIDC-IDRI	Deep CNN model	90.85	86.85	82.3	—	—
Gopi et al. (2019) [26]	LIDC-IDRI	E-CNN	97	84.7	81.7	1.7/3.8	—
Acharya et al. [33]	EPILEPSIAE	CNN	88.7	82.61	78.61	—	—
Anthimopoulos et al. [34]	ILD CT/HRCT	CNN	85.61	85.9	76.9	—	95%
Setio et al. [35]	LIDC-IDRI	ConvNet	89.9	88.7	84.7	1.0/4.0
Wang et al. [36]	DDSM	SVM	92.74	86.1	83.1	—	96.50
Liu and Tang et al. [37]	DDSM	SVM	93	87.28	84.2	—	94.39
Saki et al. [38]	MIAS	OWBPE	89.28	95.9	78.5	—	92.80