DAVS-NET: Dense Aggregation Vessel Segmentation Network for retinal vasculature detection in fundus images

Abstract

In this era, deep learning-based medical image analysis has become a reliable source in assisting medical practitioners for various retinal disease diagnosis like hypertension, diabetic retinopathy (DR), arteriosclerosis glaucoma, and macular edema etc. Among these retinal diseases, DR can lead to vision detachment in diabetic patients which cause swelling of these retinal blood vessels or even can create new vessels. This creation or the new vessels and swelling can be analyzed as biomarker for screening and analysis of DR. Deep learning-based semantic segmentation of these vessels can be an effective tool to detect changes in retinal vasculature for diagnostic purposes. This segmentation task becomes challenging because of the low-quality retinal images with different image acquisition conditions, and intensity variations. Existing retinal blood vessels segmentation methods require a large number of trainable parameters for training of their networks. This paper introduces a novel Dense Aggregation Vessel Segmentation Network (DAVS-Net), which can achieve high segmentation performance with only a few trainable parameters. For faster convergence, this network uses an encoder-decoder framework in which edge information is transferred from the first layers of the encoder to the last layer of the decoder. Performance of the proposed network is evaluated on publicly available retinal blood vessels datasets of DRIVE, CHASE_DB1, and STARE. Proposed method achieved state-of-the-art segmentation accuracy using a few number of trainable parameters.

Retinal Image Quality Assessment Using Morphological Operations

“…A total of 300 images of ROP Stage III were correctly classified by both the ophthalmologist and the algorithm to be of poor quality, while two (2) images of ROP Stage III were classified by the ophthalmologist to be of quality, but the algorithm classified them as of non-quality. Studies applying morphological operations for retina image quality assessment were divided into three categories: Image vessel extraction and feature extraction to determine available features [14], [15], [21], quality assessment studies based on assigned values [16,18,19,22,24] and image contents with edge detection [17], [20], [25]. It is observable that all studies evaluated image quality based on some parameters of image contents which constituted the available features without paying more focus on image clarity which our algorithm was able to do.…”

“…Anisotropic approach was used to normalize arrays for the vessels achieving an accuracy of 95%. Raza et al [19] developed an algorithm to check image content and reduce image noise for vessel visibility. This method was effective in detecting image edges through a developed image frame.…”

“…Abdushkour [14] 2017 Feature extraction None -Koh et al [15] 2017 Feature extraction None -Yang et al [16] 2018 Quality assessment None -Shao et al [17] 2018 Image contents None -Abdel et al [24] 2018 Image contents None -Kumar and Suraj [38] 2019 Quality assessment CLAHE -Soomro et al [18] 2018 Image content PCA 95% Raza et al [19] 2021 Image content None -Deledalle et al [20] 2019 Vessel extraction None -Rocha and Douglas [21] 2020 Quality assessment None -Wang et al [22] 2019 Quality assessment None -Zago et al [23] 2018 Quality assessment None - [25] 2023 NIQE-PIQE -97.7 % 95.6% 99% -Shi et al [26] 2022 RMHAS 220 ---0.91 Wan et al [27] 2022 CNN -96% ---Sinha [40] 2022 CNN -96% ---Guo et al [28] 2024 DNN 55931 89.7% ---Huang et al [29] 2024 RIQA 3000 -% --0.96 Bouris et al [30] 2023 CNN 32377 91% --0.98 Peter et al [31] 2024 CNN 36133 82.66% ---Qi et al [32] 2023 CNN -----Yue et al [33] 2023 CNN 2000 ----Li et al [34] 2019 SVM -----Feng et al [35] 2020 CNN -----Zhou et al [36] 2020 CNN -----Mondal et al [37] 2020 CNN -----Zhang et al [41] 2020 CNN -----Zhu and Schaefer, [42] 2020 CNN -----Tchinda et al [43] 2021 CNN -----Lenskiy et al [44] 2021 CNN -----Villalobos et al [45] 2019 CNN -----Chanwimaluang and Fan [46] 2019 CNN -----Kumar et al [38] 2019 CNN -----…”

A Systematic Study of Deep Learning Architectures for Analysis of Glaucoma and Hypertensive Retinopathy

“…The relationships between AA, VA, age, systolic blood pressure (SBP), and diastolic blood pressure were carefully observed. Authors in [16] have suggested a Dense Accumulation Vessel Segmentation Network (DAVS-Net), which can provide good segmentation accuracy with only a few trainable parameters. This network employs an encoderdecoder framework, where edge information is passed from the initial layers of the encoder to the final layer of the decoder, in order to achieve faster convergence.…”

LUVS-Net: A Lightweight U-Net Vessel Segmentor for Retinal Vasculature Detection in Fundus Images

“…Deep-learning-based techniques have shown very promising results when it comes to detecting complex patterns in a wide variety of medical images generally including those in retinal fundus images [17][18][19]. Numerous procedures to segment retinal blood vessels automatically from digital fundus images have been presented in the literature [20].…”