Table 4.

State-of-the-art CAD systems for lung cancer diagnosis.

Methods	Task Performed	Dataset	Evaluation Matrix (%)	Year	Ref.
SVM algorithm	Segmentation	Private	Acc = 89.5	2016	[100]
3D CNN trained on weakly labeled data	Nodule Detection	SPIE-LUNGx	SN = 80	2016	[101]
DCNN	Lung cancer detection	Kaggle, LUNA16	Acc = 0.75, SN = 0.77, SP = 0.74	2017	[102]
Deep residual networks	Nodule classification	LIDC/IDRI	Acc = 89.9, SN = 91, SP = 88.6	2017	[103]
3D-CNN	Detection and Classification	Bowl 2017	Acc = 86.6	2017	[104]
Polygon approximation with SVM	Nodule detection	LIDC	Acc = 98.8, SN = 97.7, SP = 96.2	2018	[90]
Deep residual networks	Nodule classification	LIDC-IDRI	Acc = 0.89, SN = 0.91, SP = 0.88	2017	[103]
Deep learning	Nodule detection	LIDC-IDRI	Acc = 0.96, SN = 0.95, SP = 0.97	2020	[105]
Deep reinforcement learning	Nodule detection	LIDC-IDRI	Acc = 0.64, SN = 0.58, SP = 0.55	2018	[106]
3D nodule candidate	Nodule detection	LIDC	Acc = 0.99, SN = 0.98, SP = 0.98	2019	[107]
Optimized Random Forest	Automatic detection	LIDC-IDRI	Acc = 93.1, SN = 94.8, SP = 91.3, FP = 0.086	2020	[91]
CNN	Segments nodules	LIDC	Acc = 89.8, SN = 85.2, SP = 90.6	2020	[108]
2D DCNN	Nodule detection	LUNA16	SN = 86.42, FP = 73.4	2019	[98]
Generative adversarial networks with DCNN	Nodule classification	Private	SN = 93.9, SP = 77.8	2020	[109]
Patch-Based CNN	Nodule detection	LIDC-IDRI	SN = 92.8	2019	[110]
SVM	Detection and segmentation	Private	SN = 90.6, SP = 73.6	2021	[111]
VGG-16 based CNN	Classifcation	Massachusetts General Hospital (MGH)	Acc = 68.6, SN = 37.5, SP = 82.9, AUC = 0.70	2021	[112]