Artificial Intelligence-Assisted Diagnosis Technology and Its Advance Based on Glaucoma Imaging

Mingyuan Li; Fengzhou Fang

doi:10.3788/LOP232292

Laser & Optoelectronics Progress, Volume. 61, Issue 14, 1400002(2024)

Artificial Intelligence-Assisted Diagnosis Technology and Its Advance Based on Glaucoma Imaging

Mingyuan Li and Fengzhou Fang^*

Laboratory of Micro/Nano Manufacturing Technology, State Key Laboratory of Precision Measuring Technology and Instruments, Tianjin University, Tianjin 300072, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(10)

Fig. 1. Number of articles published in the field of AI assisted glaucoma diagnosis in the past 5 years

Download full size

Fig. 2. Inspection sample obtained by the Humphrey perimeter^[26]

Download full size

Fig. 3. Cup and disc region map in fundus image^[30]

Download full size

Fig. 4. Flow chart of texture-based fundus image feature extraction^[36]

Download full size

Fig. 5. Sample images of OCT inspection results^[52]

Download full size

Fig. 6. OCT image around the optic nerve papilla^[55]

Download full size

Fig. 7. Schematic of FusionNet based on the combination of visual field and OCT images^[68]

Download full size

Table 1. Comparison of subtype classification study results^[28]
View table
Table 1. Comparison of subtype classification study results^[28]
Parameter Prototype method MD CIGTS AGIS PoPLR
Consistence 0.51 0.21 0.24 0.06 0.26
Accuracy 0.77 0.57 0.59 0.52 0.60

Table 2. Summary of results of detection methods based on fundus images

View table

Table 2. Summary of results of detection methods based on fundus images

Network	Dataset	AUC	Accuracy	Specificity	Sensitivity
SLIC^［42］	RIM-ONE	0.91	0.99	0.92	0.98
VGG-19^［43］	DRISHTI-GS，RIM-ONE，ESPERANZA	0.94（Total）	0.88	0.89	0.87
U-net++^［44］	ORIGA	0.913	0.843	0.793	0.894
U-net++^［44］	DRISHTI-GS1	0.912	0.837	0.772	0.904
Xception^［45］	HRF	0.8354	0.8000	0.7778	0.8333
	DRISHTI-GS1	0.8041	0.7525	0.7143	0.7419
	RIM-ONE	0.8575	0.7121	0.7990	0.7931
	HRF	0.7739	0.7082	0.7030	0.7033
	ACRIMA	0.7678	0.7021	0.7020	0.6893
TCNN，SSCNN， SSCNN-DAE^［46］	RIM-ONE（TCNN）	0.923（TCNN）	0.915	0.920	0.905
	RIGA（SSCNN，	0.936（SSCNN）	0.924	0.933	0.917
	SSCNN-DAE）	0.95（SSCNN-DAE）	0.938	0.905	0.989
CNN^［47］	RIM-ONE	0.916	0.852	0.855	0.848
CNN^［47］	LAG	0.983	0.962	0.967	0.954
DENet^［48］	SCES	0.9183	0.8429	0.8380	0.8478
DENet^［48］	SINDI	0.8173	0.7495	0.7115	0.7876

Table 3. Summary of detection methods based on OCT images

View table

Table 3. Summary of detection methods based on OCT images

Network architecture	Data type	AUROC	Accuracy	Sensitivity	Specificity
CNN，Transfer-learning^［60］	RNFL thickness，GCIPL thickness	0.937		0.825	0.939
AlexNet^［61］	RNFL thickness，GCIPL thickness	0.97	0.892
CNN^［62］	OCT B-scan	0.97			0.95
Semi-supervised CNN^［63］	OCT B-scan	0.979	0.929
ResNet^［64］	3D OCT circular scanning	0.94
VGG-16，Transfer-learning ^［65］	AS-OCT	0.90		0.79	0.87

Tools

Get Citation

Copy Citation Text

Mingyuan Li, Fengzhou Fang. Artificial Intelligence-Assisted Diagnosis Technology and Its Advance Based on Glaucoma Imaging[J]. Laser & Optoelectronics Progress, 2024, 61(14): 1400002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites