GSAC-UFormer: Groupwise Self-Attention Convolutional Transformer-Based UNet for Medical Image Segmentation

Cognitive Computation, Год журнала: 2025, Номер 17(2)

Опубликована: Фев. 22, 2025

Язык: Английский

---

MLFA-UNet: A Multi-Level Feature Assembly UNet for Medical Image Segmentation

Methods, Год журнала: 2024, Номер 232, С. 52 - 64

Опубликована: Окт. 30, 2024

Язык: Английский

---

MLFE‐UNet: Multi‐Level Feature Extraction Transformer‐Based UNet for Gastrointestinal Disease Segmentation

International Journal of Imaging Systems and Technology, Год журнала: 2025, Номер 35(1)

Опубликована: Янв. 1, 2025

ABSTRACT Accurately segmenting gastrointestinal (GI) disease regions from Wireless Capsule Endoscopy images is essential for clinical diagnosis and survival prediction. However, challenges arise due to similar intensity distributions, variable lesion shapes, fuzzy boundaries. In this paper, we propose MLFE‐UNet, an advanced fusion of CNN‐based transformers with UNet. Both the encoder decoder utilize a multi‐level feature extraction (MLFA) CNN‐Transformer‐based module. This module extracts features input data, considering both global dependencies local information. Furthermore, introduce spatial attention (MLSA) block that functions as bottleneck. It enhances network's ability handle complex structures overlapping in maps. The MLSA captures multiscale tokens channel perspective transmits them decoding path. A contextual stabilization follows each transition emulate zones facilitate segmentation guidelines at phase. To address high‐level semantic information, incorporate computationally efficient block. followed by skip connections, ensuring interaction highlighting important decoder. evaluate performance our proposed selected common GI diseases, specifically bleeding polyps. dice coefficient scores obtained MLFE‐UNet on MICCAI 2017 (Red lesion) CVC‐ClinicalDB data sets are 92.34% 88.37%, respectively.

Язык: Английский

---

Multi-Scale Cascaded Spatial Segmentation Transformer for Liver Cancer Classification

Research Square (Research Square), Год журнала: 2025, Номер unknown

Опубликована: Апрель 10, 2025

Early and accurate detection is crucial for treating liver cancer, the main cause of cancer deaths. Despite its widespread use, Computed Tomography (CT) imaging generally struggles with tumors' low contrast, uneven borders, overlapping features. The variety in tumor forms, sizes, complicated anatomical aspects makes CT image segmentation categorization difficult. Variability size shape, structures, complex anatomy are some difficulties that this method aims to address when using images diagnose cancer. Multi-Scale Cascaded Spatial Segmentation Transformer (M-SCSST) an innovative approach developed Classification Liver Cancer from Images introduced research. M-SCSST uses a cascaded processing include multi-scale spatial information into transformer-based architecture. Accurate classification heterogeneous cancers made possible by enhancing subtle features utilizing advanced attention mechanisms (AAM). Improved diagnostic accuracy achieved employing suggested on large dataset Its use helps radiologists identify cancerous benign areas, which leads earlier diagnosis better treatment choices. effectiveness scans assessed through comprehensive simulation Research measures precision, recall, computational efficiency, noise resilience, accuracy. With improved reliability, detects more effectively than conventional approaches.

Язык: Английский

---

DMFC-UFormer: Depthwise multi-scale factorized convolution transformer-based UNet for medical image segmentation

Biomedical Signal Processing and Control, Год журнала: 2024, Номер 101, С. 107200 - 107200

Опубликована: Ноя. 26, 2024

Язык: Английский

---