Multi-Scale Feature Fusion with Adaptive Weighting for Diabetic Retinopathy Severity Classification
Diabetic retinopathy (DR) is the prime cause of blindness in people who suffer from diabetes. Automation of DR diagnosis could help a lot of patients avoid the risk of blindness by identifying the disease and making judgments at an early stage. The main focus of the present work is to propose a feas...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-06-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/10/12/1369 |
| _version_ | 1797531025428971520 |
|---|---|
| author | Runze Fan Yuhong Liu Rongfen Zhang |
| author_facet | Runze Fan Yuhong Liu Rongfen Zhang |
| author_sort | Runze Fan |
| collection | DOAJ |
| description | Diabetic retinopathy (DR) is the prime cause of blindness in people who suffer from diabetes. Automation of DR diagnosis could help a lot of patients avoid the risk of blindness by identifying the disease and making judgments at an early stage. The main focus of the present work is to propose a feasible scheme of DR severity level detection under the MobileNetV3 backbone network based on a multi-scale feature of the retinal fundus image and improve the classification performance of the model. Firstly, a special residual attention module RCAM for multi-scale feature extraction from different convolution layers was designed. Then, the feature fusion by an innovative operation of adaptive weighting was carried out in each layer. The corresponding weight of the convolution block is updated in the model training automatically, with further global average pooling (GAP) and division process to avoid over-fitting of the model and removing non-critical features. In addition, Focal Loss is used as a loss function due to the data imbalance of DR images. The experimental results based on Kaggle APTOS 2019 contest dataset show that our proposed method for DR severity classification achieves an accuracy of 85.32%, a kappa statistic of 77.26%, and an AUC of 0.97. The comparison results also indicate that the model obtained is superior to the existing models and presents superior classification performance on the dataset. |
| first_indexed | 2024-03-10T10:38:15Z |
| format | Article |
| id | doaj.art-9d71056acab14f3398003d3701da8f68 |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-10T10:38:15Z |
| publishDate | 2021-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-9d71056acab14f3398003d3701da8f682023-11-21T23:10:56ZengMDPI AGElectronics2079-92922021-06-011012136910.3390/electronics10121369Multi-Scale Feature Fusion with Adaptive Weighting for Diabetic Retinopathy Severity ClassificationRunze Fan0Yuhong Liu1Rongfen Zhang2College of Big Data and Information Engineering, Guizhou University, Guiyang 550025, ChinaCollege of Big Data and Information Engineering, Guizhou University, Guiyang 550025, ChinaCollege of Big Data and Information Engineering, Guizhou University, Guiyang 550025, ChinaDiabetic retinopathy (DR) is the prime cause of blindness in people who suffer from diabetes. Automation of DR diagnosis could help a lot of patients avoid the risk of blindness by identifying the disease and making judgments at an early stage. The main focus of the present work is to propose a feasible scheme of DR severity level detection under the MobileNetV3 backbone network based on a multi-scale feature of the retinal fundus image and improve the classification performance of the model. Firstly, a special residual attention module RCAM for multi-scale feature extraction from different convolution layers was designed. Then, the feature fusion by an innovative operation of adaptive weighting was carried out in each layer. The corresponding weight of the convolution block is updated in the model training automatically, with further global average pooling (GAP) and division process to avoid over-fitting of the model and removing non-critical features. In addition, Focal Loss is used as a loss function due to the data imbalance of DR images. The experimental results based on Kaggle APTOS 2019 contest dataset show that our proposed method for DR severity classification achieves an accuracy of 85.32%, a kappa statistic of 77.26%, and an AUC of 0.97. The comparison results also indicate that the model obtained is superior to the existing models and presents superior classification performance on the dataset.https://www.mdpi.com/2079-9292/10/12/1369diabetic retinopathymulti-scale featureattention moduleadaptive weightingfeature fusion |
| spellingShingle | Runze Fan Yuhong Liu Rongfen Zhang Multi-Scale Feature Fusion with Adaptive Weighting for Diabetic Retinopathy Severity Classification Electronics diabetic retinopathy multi-scale feature attention module adaptive weighting feature fusion |
| title | Multi-Scale Feature Fusion with Adaptive Weighting for Diabetic Retinopathy Severity Classification |
| title_full | Multi-Scale Feature Fusion with Adaptive Weighting for Diabetic Retinopathy Severity Classification |
| title_fullStr | Multi-Scale Feature Fusion with Adaptive Weighting for Diabetic Retinopathy Severity Classification |
| title_full_unstemmed | Multi-Scale Feature Fusion with Adaptive Weighting for Diabetic Retinopathy Severity Classification |
| title_short | Multi-Scale Feature Fusion with Adaptive Weighting for Diabetic Retinopathy Severity Classification |
| title_sort | multi scale feature fusion with adaptive weighting for diabetic retinopathy severity classification |
| topic | diabetic retinopathy multi-scale feature attention module adaptive weighting feature fusion |
| url | https://www.mdpi.com/2079-9292/10/12/1369 |
| work_keys_str_mv | AT runzefan multiscalefeaturefusionwithadaptiveweightingfordiabeticretinopathyseverityclassification AT yuhongliu multiscalefeaturefusionwithadaptiveweightingfordiabeticretinopathyseverityclassification AT rongfenzhang multiscalefeaturefusionwithadaptiveweightingfordiabeticretinopathyseverityclassification |