Multi-module attention-guided deep learning framework for precise gastrointestinal disease identification in endoscopic imagery

Biomedical Signal Processing and Control, Год журнала: 2024, Номер 95, С. 106396 - 106396

Опубликована: Май 1, 2024

Язык: Английский

---

Unveiling camouflaged and partially occluded colorectal polyps: Introducing CPSNet for accurate colon polyp segmentation

Computers in Biology and Medicine, Год журнала: 2024, Номер 171, С. 108186 - 108186

Опубликована: Фев. 21, 2024

Язык: Английский

---

UViT-Seg: An Efficient ViT and U-Net-Based Framework for Accurate Colorectal Polyp Segmentation in Colonoscopy and WCE Images

Deleted Journal, Год журнала: 2024, Номер 37(5), С. 2354 - 2374

Опубликована: Апрель 26, 2024

Colorectal cancer (CRC) stands out as one of the most prevalent global cancers. The accurate localization colorectal polyps in endoscopy images is pivotal for timely detection and removal, contributing significantly to CRC prevention. manual analysis generated by gastrointestinal screening technologies poses a tedious task doctors. Therefore, computer vision-assisted could serve an efficient tool polyp segmentation. Numerous efforts have been dedicated automating localization, with majority studies relying on convolutional neural networks (CNNs) learn features from images. Despite their success segmentation tasks, CNNs exhibit significant limitations precisely determining location shape due sole reliance learning local While manifest variation features, encompassing both high- low-level ones, framework that combines ability desired. This paper introduces UViT-Seg, designed Operating encoder-decoder architecture, UViT-Seg employs two distinct feature extraction methods. A vision transformer encoder section captures long-range semantic information, while CNN module, integrating squeeze-excitation dual attention mechanisms, focusing critical image regions. Experimental evaluations conducted five public datasets, including CVC clinic, ColonDB, Kvasir-SEG, ETIS LaribDB, Kvasir Capsule-SEG, demonstrate UViT-Seg's effectiveness localization. To confirm its generalization performance, model tested datasets not used training. Benchmarking against common methods state-of-the-art approaches, proposed yields promising results. For instance, it achieves mean Dice coefficient 0.915 intersection over union 0.902 Colon dataset. Furthermore, has advantage being efficient, requiring fewer computational resources training testing. positions optimal choice real-world deployment scenarios.

Язык: Английский

---

An innovative transfer learning-polyp detection from wireless capsule endoscopy videos with optimal key frame selection and depth estimation

Biomedical Signal Processing and Control, Год журнала: 2025, Номер 108, С. 107963 - 107963

Опубликована: Май 3, 2025

Язык: Английский

---

Multi-stage feature fusion network for polyp segmentation

Applied Soft Computing, Год журнала: 2025, Номер unknown, С. 113034 - 113034

Опубликована: Март 1, 2025

Язык: Английский

---

Distributed swarm learning for communication-efficient capsule endoscopy classification on solving Non-i.i.d data

Alexandria Engineering Journal, Год журнала: 2025, Номер 128, С. 92 - 99

Опубликована: Май 26, 2025

Язык: Английский

---

SwinSAM: Fine-grained polyp segmentation in colonoscopy images via segment anything model integrated with a Swin Transformer decoder

Biomedical Signal Processing and Control, Год журнала: 2024, Номер 100, С. 107055 - 107055

Опубликована: Ноя. 15, 2024

Язык: Английский

---

DHAFormer: Dual-channel hybrid attention network with transformer for polyp segmentation

PLoS ONE, Год журнала: 2024, Номер 19(7), С. e0306596 - e0306596

Опубликована: Июль 10, 2024

The accurate early diagnosis of colorectal cancer significantly relies on the precise segmentation polyps in medical images. Current convolution-based and transformer-based methods show promise but still struggle with varied sizes shapes often low contrast between their background. This research introduces an innovative approach to confronting aforementioned challenges by proposing a Dual-Channel Hybrid Attention Network Transformer (DHAFormer). Our proposed framework features multi-scale channel fusion module, which excels at recognizing across spectrum shapes. Additionally, framework’s dual-channel hybrid attention mechanism is innovatively conceived reduce background interference improve foreground representation polyp integrating local global information. DHAFormer demonstrates significant improvements task compared currently established methodologies.

Язык: Английский

---

BCL-Former: Localized Transformer Fusion with Balanced Constraint for polyp image segmentation

Computers in Biology and Medicine, Год журнала: 2024, Номер 182, С. 109182 - 109182

Опубликована: Сен. 27, 2024

Язык: Английский

---

PFPRNet: A Phase-Wise Feature Pyramid With Retention Network for Polyp Segmentation

IEEE Journal of Biomedical and Health Informatics, Год журнала: 2024, Номер 29(2), С. 1137 - 1150

Опубликована: Ноя. 18, 2024

Early detection of colonic polyps is crucial for the prevention and diagnosis colorectal cancer. Currently, deep learning-based polyp segmentation methods have become mainstream achieved remarkable results. Acquiring a large number labeled data time-consuming labor-intensive, meanwhile presence numerous similar wrinkles in images also hampers model prediction performance. In this paper, we propose novel approach called Phase- wise Feature Pyramid with Retention Network (PFPRNet), which leverages pre-trained Transformer-based Encoder to obtain multi-scale feature maps. A Decoder designed gradually integrate global features into local guide model's attention towards key regions. Additionally, our custom Enhance Perception module enables capturing image information from broader perspective. Finally, introduce an innovative Low-layer as alternative Transformer more efficient modeling. Evaluation results on several widely-used datasets demonstrate that proposed method has strong learning ability generalization capability, outperforms state-of-the-art approaches.

Язык: Английский

---