Table 6 Comparison of results with preprocessing and data augmentation.

From: Attention LinkNet-152: a novel encoder-decoder based deep learning network for automated spine segmentation

Methodology

Dice Coefficient (DSC) (%)

Jaccard Index (IoU) (%)

Training Loss (%)

Validation Loss (%)

Preprocessing

93.5

91.2

0.004

0.005

Data Augmentation

94.7

92.5

0.003

0.004

Back to article page