A retinal vessel segmentation network with multiple-dimension attention and adaptive feature fusion

Computers in Biology and Medicine, Journal Year: 2024, Volume and Issue: 172, P. 108315 - 108315

Published: March 15, 2024

Language: Английский

---

VSR-Net: Vessel-like Structure Rehabilitation Network with Graph Clustering

IEEE Transactions on Image Processing, Journal Year: 2025, Volume and Issue: 34, P. 1090 - 1105

Published: Jan. 1, 2025

The morphologies of vessel-like structures, such as blood vessels and nerve fibres, play significant roles in disease diagnosis, e.g., Parkinson's disease. Although deep network-based refinement segmentation topology-preserving methods recently have achieved promising results segmenting they still face two challenges: (1) existing often limitations rehabilitating subsection ruptures segmented structures; (2) are typically overconfident predicted results. To tackle these challenges, this paper attempts to leverage the potential spatial interconnection relationships among from structure rehabilitation perspective. Based on perspective, we propose a novel Vessel-like Structure Rehabilitation Network (VSR-Net) both rehabilitate improve model calibration based coarse VSR-Net first constructs rupture clusters via Curvilinear Clustering Module (CCM). Then, well-designed Merging (CMM) is applied obtain refined structures. Extensive experiments six 2D/3D medical image datasets show that significantly outperforms state-of-the-art (SOTA) with lower errors. Additionally, provide quantitative analysis explain morphological difference between VSR-Net's ground truth (GT), which smaller compared those SOTA GT, demonstrating our method more effectively rehabilitates

Language: Английский

---

Attention-guided cascaded network with pixel-importance-balance loss for retinal vessel segmentation

Frontiers in Cell and Developmental Biology, Journal Year: 2023, Volume and Issue: 11

Published: May 9, 2023

Accurate retinal vessel segmentation from fundus images is essential for eye disease diagnosis. Many deep learning methods have shown great performance in this task but still struggle with limited annotated data. To alleviate issue, we propose an Attention-Guided Cascaded Network (AGC-Net) that learns more valuable features a few images. Attention-guided cascaded network consists of two stages: the coarse stage produces rough prediction map image, and fine refines missing details map. In attention-guided network, incorporate inter-stage attention module (ISAM) to cascade backbone these stages, which helps focus on regions better refinement. We also Pixel-Importance-Balance Loss (PIB Loss) train model, avoids gradient domination by non-vascular pixels during backpropagation. evaluate our mainstream image datasets (i.e., DRIVE CHASE-DB1) achieve AUCs 0.9882 0.9914, respectively. Experimental results show method outperforms other state-of-the-art performance.

Language: Английский

---

An evolutionary U-shaped network for Retinal Vessel Segmentation using Binary Teaching–Learning-Based Optimization

Biomedical Signal Processing and Control, Journal Year: 2023, Volume and Issue: 83, P. 104669 - 104669

Published: Feb. 9, 2023

Language: Английский

---

Dual-Path and Multi-Scale Enhanced Attention Network for Retinal Diseases Classification Using Ultra-Wide-Field Images

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 45405 - 45415

Published: Jan. 1, 2023

Early computer-aided early diagnosis (CAD) based on retinal imaging is critical to the timely management and treatment planning of retina-related diseases. However, inherent characteristics images complexity their pathological patterns, such as low image contrast different lesion sizes, restrict performance CAD systems. Recently, ultra-wide-field (UWF) have become a useful tool for disease detection due capability capturing much broader view retina (i.e., up 200°), in comparison with most commonly used fundus (45°). In this paper, we propose an attention-based multi-branch network diseases classification four subject groups. The proposed method consists multi-scale feature fusion module dual attention module. Specifically, small-scale lesions are identified using features extracted from To better explore obtained features, global graph incorporated enable recognize salient objects interest. Comprehensive validations both private public datasets were carried out verify effectiveness model.

Language: Английский

---

Metrics for comparison of image dataset and segmentation methods for fractal analysis of retinal vasculature

Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 105, P. 107650 - 107650

Published: Feb. 12, 2025

Language: Английский

---

Collation of a Few Retinal Vessel Segmentation Techniques: Is the Problem Solved?

SN Computer Science, Journal Year: 2025, Volume and Issue: 6(2)

Published: Feb. 13, 2025

Language: Английский

---

SAMP-Net: a medical image segmentation network with split attention and multi-layer perceptron

Medical & Biological Engineering & Computing, Journal Year: 2025, Volume and Issue: unknown

Published: March 11, 2025

Language: Английский

---

Exploring a multi-path U-net with probability distribution attention and cascade dilated convolution for precise retinal vessel segmentation in fundus images

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 18, 2025

Abstract While deep learning has become the go-to method for image denoising due to its impressive noise removal Retinal blood vessel segmentation presents several challenges, including limited labeled data, complex multi-scale structures, and susceptibility interference from lesion areas. To confront these this work offers a novel technique that integrates attention mechanisms cascaded dilated convolution module (CDCM) within multi-path U-Net architecture. First, dual-path is developed extract both coarse fine-grained structures through separate texture structural branches. A CDCM integrated gather features, enhancing model’s ability semantic features. Second, boosting algorithm incorporates probability distribution (PDA) upscaling blocks employed. This approach adjusts distribution, increasing contribution of shallow information, thereby performance in backgrounds reducing risk overfitting. Finally, output processed feature refinement module. step further refines by integrating extracting relevant Results experiments on three benchmark datasets, CHASEDB1, DRIVE, STARE, demonstrate proposed delivers improved accuracy compared existing techniques.

Language: Английский

---

A Convolutional Autoencoder Approach for Boosting the Specificity of Retinal Blood Vessels Segmentation

Applied Sciences, Journal Year: 2023, Volume and Issue: 13(5), P. 3255 - 3255

Published: March 3, 2023

Automated retina vessel segmentation of the human eye plays a vital role as it can significantly assist ophthalmologists in identifying many diseases, such diabetes, stroke, arteriosclerosis, cardiovascular disease, and other illnesses. The fast, automatic accurate eyes is very desirable. This paper introduces novel fully convolutional autoencoder for task. proposed model consists eight layers, each consisting convolutional2D MaxPooling Batch Normalisation layers more. Our has been trained evaluated on DRIVE STARE datasets with 35 min training time. performance we introduce assessed two public datasets, achieved quite competitive results compared to state-of-the-art methods literature. In particular, our reached an accuracy 95.73, AUC_ROC 97.49 dataset, 96.92 AUC ROC 97.57 dataset. Furthermore, demonstrated highest specificity among literature, reporting 98.57 98.7 respectively. above statement be noticed final blood images produced by method since segmentations are more accurate, sharp noiseless than result methods.

Language: Английский

---