Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Region Guided Attention Network for Retinal Vessel Segmentation

Syed Javed, Tariq M. Khan, Abdul Qayyum, Arcot Sowmya, Imran Razzak

TL;DR

A lightweight retinal vessel segmentation network based on the encoder-decoder mechanism with region-guided attention with weighted dice loss is presented, encouraging the model to generate more accurate segmentation with improved object boundary delineation and reduced fragmentation.

Abstract

Retinal imaging has emerged as a promising method of addressing this challenge, taking advantage of the unique structure of the retina. The retina is an embryonic extension of the central nervous system, providing a direct in vivo window into neurological health. Recent studies have shown that specific structural changes in retinal vessels can not only serve as early indicators of various diseases but also help to understand disease progression. In this work, we present a lightweight retinal vessel segmentation network based on the encoder-decoder mechanism with region-guided attention. We introduce inverse addition attention blocks with region guided attention to focus on the foreground regions and improve the segmentation of regions of interest. To further boost the model's performance on retinal vessel segmentation, we employ a weighted dice loss. This choice is particularly effective in addressing the class imbalance issues frequently encountered in retinal vessel segmentation tasks. Dice loss penalises false positives and false negatives equally, encouraging the model to generate more accurate segmentation with improved object boundary delineation and reduced fragmentation. Extensive experiments on a benchmark dataset show better performance (0.8285, 0.8098, 0.9677, and 0.8166 recall, precision, accuracy and F1 score respectively) compared to state-of-the-art methods.

Region Guided Attention Network for Retinal Vessel Segmentation

TL;DR

A lightweight retinal vessel segmentation network based on the encoder-decoder mechanism with region-guided attention with weighted dice loss is presented, encouraging the model to generate more accurate segmentation with improved object boundary delineation and reduced fragmentation.

Abstract

Retinal imaging has emerged as a promising method of addressing this challenge, taking advantage of the unique structure of the retina. The retina is an embryonic extension of the central nervous system, providing a direct in vivo window into neurological health. Recent studies have shown that specific structural changes in retinal vessels can not only serve as early indicators of various diseases but also help to understand disease progression. In this work, we present a lightweight retinal vessel segmentation network based on the encoder-decoder mechanism with region-guided attention. We introduce inverse addition attention blocks with region guided attention to focus on the foreground regions and improve the segmentation of regions of interest. To further boost the model's performance on retinal vessel segmentation, we employ a weighted dice loss. This choice is particularly effective in addressing the class imbalance issues frequently encountered in retinal vessel segmentation tasks. Dice loss penalises false positives and false negatives equally, encouraging the model to generate more accurate segmentation with improved object boundary delineation and reduced fragmentation. Extensive experiments on a benchmark dataset show better performance (0.8285, 0.8098, 0.9677, and 0.8166 recall, precision, accuracy and F1 score respectively) compared to state-of-the-art methods.
Paper Structure (15 sections, 8 equations, 3 figures, 5 tables)

This paper contains 15 sections, 8 equations, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Light-weight Models for Medical Image Segmentation
  4. Region Guided Attention Network
  5. Encoder Block
  6. Decoder Block
  7. Inverse Addition Attention Block
  8. Proposed Network Architecture
  9. Results and Discussion
  10. Dataset
  11. Implementation Details
  12. Ablation Study
  13. Comparison with Existing Works
  14. Qualitative Results
  15. Conclusion and Future Work

Figures (3)

  • Figure 1: Block diagram of the proposed methodology. The figure depicts the overall proposed methodology, detailing the Encoder, Bottleneck, and Decoder Blocks. The detailed layers and operations of the Partial Decoder and Inverse Addition Attention blocks are given in the respective sections. The figure also explains what each symbol in the flow chart means.
  • Figure 2: Qualitative results of the proposed method on some sample images from the DRIVE DRIVEdata dataset. The columns from left to right show the query image, segmentation mask (ground truth), and the mask predicted by the model and analytic mask respectively. The green pixels in the analytic mask represent the correctly segmented pixels while the red pixels are the false negatives and the blue pixels are the false positives.
  • Figure 3: Qualitative results of the proposed method on some sample images from the CHASEDB1 CHASEDataset dataset. The columns from left to right show the query image, segmentation mask (ground truth), and the mask predicted by the model and analytic mask respectively. The green pixels in the analytic mask represent the correctly segmented pixels while the red pixels are the false negatives and the blue pixels are the false positives.