Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation Method
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...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-03-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/23/7/3431 |
| _version_ | 1797607070616256512 |
|---|---|
| author | Vidas Raudonis Arturas Kairys Rasa Verkauskiene Jelizaveta Sokolovska Goran Petrovski Vilma Jurate Balciuniene Vallo Volke |
| author_facet | Vidas Raudonis Arturas Kairys Rasa Verkauskiene Jelizaveta Sokolovska Goran Petrovski Vilma Jurate Balciuniene Vallo Volke |
| author_sort | Vidas Raudonis |
| collection | DOAJ |
| description | 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 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>I</mi><mi>o</mi><mi>U</mi></mrow></semantics></math></inline-formula> values. The ensemble-based model achieved higher Dice score (0.95) and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>I</mi><mi>o</mi><mi>U</mi></mrow></semantics></math></inline-formula> (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. |
| first_indexed | 2024-03-11T05:26:07Z |
| format | Article |
| id | doaj.art-b9aad7cffc8f44ecba4fe5a2b3dd1c0b |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-11T05:26:07Z |
| publishDate | 2023-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-b9aad7cffc8f44ecba4fe5a2b3dd1c0b2023-11-17T17:32:26ZengMDPI AGSensors1424-82202023-03-01237343110.3390/s23073431Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation MethodVidas Raudonis0Arturas Kairys1Rasa Verkauskiene2Jelizaveta Sokolovska3Goran Petrovski4Vilma Jurate Balciuniene5Vallo Volke6Automation Department, Kaunas University of Technology, 51368 Kaunas, LithuaniaAutomation Department, Kaunas University of Technology, 51368 Kaunas, LithuaniaInstitute of Endocrinology, Lithuanian University of Health Sciences, 50140 Kaunas, LithuaniaFaculty of Medicine, University of Latvia, 1004 Riga, LatviaCenter of Eye Research and Innovative Diagnostics, Department of Ophthalmology, Oslo University Hospital and Institute of Clinical Medicine, Faculty of Medicine, University of Oslo, 0372 Oslo, NorwayLithuanian University of Health Sciences, 44307 Kaunas, LithuaniaFaculty of Medicine, Tartu University, 50411 Tartu, EstoniaIn 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 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>I</mi><mi>o</mi><mi>U</mi></mrow></semantics></math></inline-formula> values. The ensemble-based model achieved higher Dice score (0.95) and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>I</mi><mi>o</mi><mi>U</mi></mrow></semantics></math></inline-formula> (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.https://www.mdpi.com/1424-8220/23/7/3431diabetic retinopathy (DR)image segmentationmicroaneurysms (MAs)encoder-decoder deep neural network |
| spellingShingle | Vidas Raudonis Arturas Kairys Rasa Verkauskiene Jelizaveta Sokolovska Goran Petrovski Vilma Jurate Balciuniene Vallo Volke Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation Method Sensors diabetic retinopathy (DR) image segmentation microaneurysms (MAs) encoder-decoder deep neural network |
| title | Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation Method |
| title_full | Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation Method |
| title_fullStr | Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation Method |
| title_full_unstemmed | Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation Method |
| title_short | Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation Method |
| title_sort | automatic detection of microaneurysms in fundus images using an ensemble based segmentation method |
| topic | diabetic retinopathy (DR) image segmentation microaneurysms (MAs) encoder-decoder deep neural network |
| url | https://www.mdpi.com/1424-8220/23/7/3431 |
| work_keys_str_mv | AT vidasraudonis automaticdetectionofmicroaneurysmsinfundusimagesusinganensemblebasedsegmentationmethod AT arturaskairys automaticdetectionofmicroaneurysmsinfundusimagesusinganensemblebasedsegmentationmethod AT rasaverkauskiene automaticdetectionofmicroaneurysmsinfundusimagesusinganensemblebasedsegmentationmethod AT jelizavetasokolovska automaticdetectionofmicroaneurysmsinfundusimagesusinganensemblebasedsegmentationmethod AT goranpetrovski automaticdetectionofmicroaneurysmsinfundusimagesusinganensemblebasedsegmentationmethod AT vilmajuratebalciuniene automaticdetectionofmicroaneurysmsinfundusimagesusinganensemblebasedsegmentationmethod AT vallovolke automaticdetectionofmicroaneurysmsinfundusimagesusinganensemblebasedsegmentationmethod |