Table 7. Classification results of the IQ-OTH/NCCD^a and chest CT^b scan images datasets using preactivated deep learning models with various data augmentation techniques and segmentation of the lung region of interest (224×224 image resolution).

Data augmentation	Rankc	ResNet-34		MobileNetV3 (small)		Vision Transformer (base-16)		Swin Transformer (tiny)
		Accuracy, %	AUROCd, %	Accuracy, %	AUROC, %	Accuracy, %	AUROC, %	Accuracy, %	AUROC, %
IQ-OTH/NCCD dataset
Base modele	5	96.65	99.13	95.43	97.28	89.94	96.51	93.60	98.21
Cutoute	4	96.04	98.86	95.43	96.33	92.38	96.20	93.90	97.80
Random Erasinge	3	95.73	97.45	96.65f	97.29	91.46	96.27	94.82f	98.00
MixUpe,g	6	95.87	99.19f	91.77	97.11	91.77	96.27	93.29	97.52
CutMix	2	96.65	98.86	94.51	96.39	93.90f	97.64f	93.29	97.52
Random Pixel Swapg	1	97.56f	98.61	96.65f	98.00f	92.38	96.97	94.82f	98.12f
Chest CT scan images dataset
Base model	2	95.19	99.03	87.82	96.83f	82.69	95.48	90.71	98.11
Cutoute	4	94.55	98.85	88.14	97.66	80.77	93.86	88.14	97.32
Random Erasinge,g	5	94.55	98.75	86.54	96.52	79.81	89.72	86.86	97.16
MixUpe,g	6	94.55	98.77	82.05	95.33	78.85	93.29	85.90	97.10
CutMix	3	95.19	99.05f	86.54	96.89	86.86f	96.43f	87.82	96.73
Random Pixel Swap	1	95.51f	98.86	90.71f	97.51	83.65	95.83	91.35f	98.36f

^aIQ-OTH/NCCD: Iraq-Oncology Teaching Hospital/National Center for Cancer Diseases.

^bCT: computed tomography.

^cRank represents the overall rating for each technique, with “1” indicating the best technique across all models.

^dAUROC: area under the receiver operating characteristic curve.

^eSignificant difference between an augmentation technique and the Random Pixel Swap technique across all models.

^fHighest value in the column.

^gSignificant difference between training using an augmentation technique and the base model across all models.