Skip to main content

Table 3 Comparison of the results of the proposed method for retinal hemorrhage segmentation to the latest current methods in terms of similarity coefficient (Dice), intersection over union (IoU), precision and loss for training data

From: Hemorrhage semantic segmentation in fundus images for the diagnosis of diabetic retinopathy by using a convolutional neural network

Authors and Refs.

Dataset

Dice (%)

IOU (%)

Precision (%)

Loss for training data

Ananda et al. [37]

IDRID and MESSIDOR

86.53

_

_

0.1461

Proposed

IDRID and DIARETDB1

86.51

76.61

99.98

0.0038