Fundus Lesions Segmentation
Collection
Models for segmentation of fundus images
•
1 item
•
Updated
Lesions Segmentation in Fundus
Introduction
We focus on the semantic segmentations of:
- Cotton Wool Spot
- Exudates
- Hemmorrhages
- Microaneurysms
For an easier use of the models, we refer to cleaned-up version of the code provided in the fundus lesions toolkit.
Architecture
The model uses unet_seresnext50_32x4d as architecture. The implementation is taken from torchSeg
Training datasets
The model was trained on the following datasets: DDR, FGADR, IDRID, MESSIDOR, RETLES
Resolution
The image resolution for training was set to 1024 x 1024.
- Downloads last month
- 3,552
Collection including ClementP/fundus-lesions-segmentation-unet_seresnext50_32x4d
Evaluation results
- AUC Precision Recall - IDRID COTTON_WOOL_SPOT - COTTON_WOOL_SPOT on IDRIDself-reported0.670
- AUC Precision Recall - IDRID EXUDATES - EXUDATES on IDRIDself-reported0.786
- AUC Precision Recall - IDRID HEMORRHAGES - HEMORRHAGES on IDRIDself-reported0.674
- AUC Precision Recall - IDRID MICROANEURYSMS - MICROANEURYSMS on IDRIDself-reported0.398
- AUC Precision Recall - FGADR COTTON_WOOL_SPOT - COTTON_WOOL_SPOT on FGADRself-reported0.445
- AUC Precision Recall - FGADR EXUDATES - EXUDATES on FGADRself-reported0.695
- AUC Precision Recall - FGADR HEMORRHAGES - HEMORRHAGES on FGADRself-reported0.651
- AUC Precision Recall - FGADR MICROANEURYSMS - MICROANEURYSMS on FGADRself-reported0.290
- AUC Precision Recall - MESSIDOR COTTON_WOOL_SPOT - COTTON_WOOL_SPOT on MESSIDORself-reported0.331
- AUC Precision Recall - MESSIDOR EXUDATES - EXUDATES on MESSIDORself-reported0.712