Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 Mar 24;23(7):3431.
doi: 10.3390/s23073431.

Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation Method

Vidas Raudonis  1 , Arturas Kairys  1 , Rasa Verkauskiene  2 , Jelizaveta Sokolovska  3 , Goran Petrovski  4   5 , Vilma Jurate Balciuniene  6 , Vallo Volke  7
Affiliations

Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation Method

Vidas Raudonis et al. Sensors (Basel). .

Abstract

In this study, a novel method for automatic microaneurysm detection in color fundus images is presented. The proposed method is based on three main steps: (1) image breakdown to smaller image patches, (2) inference to segmentation models, and (3) reconstruction of the predicted segmentation map from output patches. The proposed segmentation method is based on an ensemble of three individual deep networks, such as U-Net, ResNet34-UNet and UNet++. The performance evaluation is based on the calculation of the Dice score and IoU values. The ensemble-based model achieved higher Dice score (0.95) and IoU (0.91) values compared to other network architectures. The proposed ensemble-based model demonstrates the high practical application potential for detection of early-stage diabetic retinopathy in color fundus images.

Keywords: diabetic retinopathy (DR); encoder-decoder deep neural network; image segmentation; microaneurysms (MAs).

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Figure 1
Figure 1
Examples of fundus images of the left and right eye with several microaneurysms.
Figure 2
Figure 2
An example of a color fundus image (left) and annotated image represented as an MA segmentation map (right).
Figure 3
Figure 3
Functional selection diagram of the region of interest based on an overlapping sliding window approach.
Figure 4
Figure 4
Deep neural network structure of the U-Net model.
Figure 5
Figure 5
The deep neural network structure of the ResNet34-UNet model.
Figure 6
Figure 6
An example of ResBlock within a residual network.
Figure 7
Figure 7
The architecture of the U-Net++ segmentation model.
Figure 8
Figure 8
Functional diagram of the proposed ensemble-based segmentation model.
Figure 9
Figure 9
Diagram of the reconstruction of the whole MA segmentation map.
Figure 10
Figure 10
Visualization of the microaneurysm (MA) segmentation results of RoIs acquired from color fundus images: (a) original RoI, (b) the corresponding ground truth annotation, (c) MA segmentation results using the U-Net model, (d) MA segmentation results using the residual U-Net model, (e) MA segmentation results using the U-Net++ model and (f) MA segmentation results using the ensemble of models.
Figure 11
Figure 11
Relationship graph between the IoU value, tested image ID and segmentation model.
Figure 12
Figure 12
Relationship graph between the Dice score, tested image ID and segmentation model.
Figure 13
Figure 13
Samples of predicted whole-image segmentation maps reconstructed from patches acquired using the ensemble-based model.
See this image and copyright information in PMC

References

    1. International Diabetes Federation . IDF Diabetes Atlas. 10th ed. International Diabetes Federation; Brussels, Belgium: 2021.
    1. GBD 2019 Blindness and Vision Impairment Collaborators, Vision Loss Expert Group of the Global Burden of Disease Study Causes of blindness and vision impairment in 2020 and trends over 30 years, and prevalence of avoidable blindness in relation to VISION 2020: The Right to Sight: An analysis for the Global Burden of Disease Study. Lancet Glob. Health. 2021;9:e144–e160. doi: 10.1016/S2214-109X(20)30489-7. Erratum in Lancet Glob. Health 2021, 9, e408. - DOI - PMC - PubMed
    1. Ehlers J.P., Jiang A.C., Boss J.D., Hu M., Figueiredo N., Babiuch A., Talcott K., Sharma S., Hach J., Le T., et al. Quantitative ultra-widefield angiography and diabetic retinopathy severity: An assessment of panretinal leakage index, ischemic index and microaneurysm count. Ophthalmology. 2019;126:1527–1532. doi: 10.1016/j.ophtha.2019.05.034. - DOI - PMC - PubMed
    1. Nadeem M.W., Goh H.G., Hussain M., Liew S.-Y., Andonovic I., Khan M.A. Deep Learning for Diabetic Retinopathy Analysis: A Review, Research Challenges, and Future Directions. Sensors. 2022;22:6780. doi: 10.3390/s22186780. - DOI - PMC - PubMed
    1. Palani D., Venkatalakshmi K., Jabeen A.R., Ram V.M.A.B. Effective Detection of Diabetic Retinopathy from Human Retinal Fundus Images Using Modified FCM and IWPSO; Proceedings of the 2019 IEEE International Conference on System, Computation, Automation and Networking (ICSCAN); Pondicherry, India. 29–30 March 2019; pp. 1–5. - DOI

MeSH terms