Diagnosis of Diabetic Retinopathy through Retinal Fundus Images and 3D Convolutional Neural Networks with Limited Number of Samples
Table 6
Class-wise IBA values for the multiclass classification tasks without augmentation, with random weak Gaussian blurred augmentation, with random shift augmentation, and with combined augmentations.
Method
IBA (mild)
IBA (moderate)
IBA (no)
IBA (proliferate)
IBA (severe)
Without augmentation
0.1366
0.0649
0.1232
0.4747
0.1183
With random weak Gaussian blurred augmentation
0.1176
0.09
0.088
0.2458
0.1048
With random shifted augmentation
0.0967
0.0631
0.0833
0.3109
0.1095
With combined augmentations
0.1203
0.1
0.0997
0.3086
0.1031
Test set validated on the model trained without augmentation