Application of Deep Learning Methods in Diagnosis of Lung Nodules

Bin Cao; Feng Yang; Jingang Ma

doi:10.3788/LOP202158.1600005

Journals >Laser & Optoelectronics Progress >Volume 58 >Issue 16 >Page 1600005 > Article

Laser & Optoelectronics Progress
Vol. 58, Issue 16, 1600005 (2021)

Application of Deep Learning Methods in Diagnosis of Lung Nodules

Bin Cao¹, Feng Yang¹, and Jingang Ma^2、*

Author Affiliations

¹Shandong Provincial Hospital of Traditional Chinese Medicine, Jinan, Shandong 250000, China

²School of Intelligence and Information Engineering, Shandong University of Traditional Chinese Medicine, Jinan, Shandong 250355, China

show less

DOI: 10.3788/LOP202158.1600005 Cite this Article Set citation alerts

Bin Cao, Feng Yang, Jingang Ma. Application of Deep Learning Methods in Diagnosis of Lung Nodules[J]. Laser & Optoelectronics Progress, 2021, 58(16): 1600005 Copy Citation Text

show less

Fig. 1. Schematic diagram of segmentation of lung nodules by U-Net

Download full size

Fig. 2. Comparison of 2D-CNN and 3D-CNN structures

Download full size

Fig. 3. Schematic diagram of Faster R-CNN lung nodule detection

Download full size

Fig. 4. Specific structure of DBN classifier

Download full size

Dataset	Image format	Number of cases	Number of CT scans	Nodule tag	Note
LIDC-IDRI^[7]	DICOM	1010	1018	*	Open
LUNA16^[8]	mhd	888	888	*	Not open
DSB^[10]	DICOM	2101	2 101	--	Not open
NLST^[11]	DICOM	--	3 410	*	Open
DLCST^[12]	DICOM	823	17	*	Open
LUNGx^[13]	DICOM	--	70	*	Open
ANODEO9^[14]	DICOM	--	55	*	Open
AliTianchi^[15]	mhd	1600	2 000	*	Not open

Table 1. Data set of lung nodule diagnosis field

Reference	Time	Dataset	CT number	FP per scan	Method	Sensitivity /%	Note
Miao et al.^[19]	2018	LIDC-IDRI	888	1.0 or 4.0	2D	87.3	2D FCN
Chen et al.^[20]	2017	DLCST	612	6	2D	76.5	CNN
Tong et al.^[22]	2018	LIDC-IDRI	888	1.0 or 20	2D	96.7 or 98.3	U-Net+ResNet
Ruan et al.^[23]	2020	LUNA16	888	1.0 or 4.0	2D	90.1	LSTM+ U-Net
Setio et al. ^[24]	2017	LUNA16	888	1.0 or 4.0	2D	85.4 or 90.1	CNN
Dou et al.^[25]	2017	LIDC-IDRI	888	219.1	3D	97.1	FCN+ResNet
Liu et al.^[26]	2017	LIDC-IDRI	888	--	3D	95.8	3D-DenseNet+U-net
Cao et al.^[29]	2020	LIDC-IDRI	888	1.0 or 4.0	3D	85.4	CNN
Hou et al.^[30]	2020	LUNA16	888	--	3D	92.3	3D U-Net+CRF
Dehmeshki et al. ^[31]	2008	LIDC-IDRI	25	--	3D	93.4	U-Net+ResNet
Xie et al. ^[33]	2019	LUNA16	888	0.125 or 0.25	3D	73.4 or 74.4	CNN
Yuan et al. ^[35]	2018	LIDC-IDRI	888	--	3D	96.7 or 98.3	CNN
Xie et al. ^[38]	2019	LUNA16	888	--	2D+3D	86.42	Faster R-CNN
Ding et al. ^[39]	2017	LUNA16	888	1.0 or 4.0	2D+3D	92.2 or 94.4	Faster R-CNN+CNN
Zheng et al. ^[40]	2020	LIDC-IDRI	888	1.0 or 2.0	2D+3D	92.67 or 94.19	U-Net+Faster R-CNN

Table 2. Comparison of results of pulmonary nodule detection models in selected literatures

Bin Cao, Feng Yang, Jingang Ma. Application of Deep Learning Methods in Diagnosis of Lung Nodules[J]. Laser & Optoelectronics Progress, 2021, 58(16): 1600005

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information