Modified U-Net with attention gate for enhanced automated brain tumor segmentation

Neural Computing and Applications, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 2, 2025

Language: Английский

---

Glaucoma detection and severity classification based on glaucoattent net framework

International Journal of Machine Learning and Cybernetics, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 28, 2025

Language: Английский

---

HDL-ACO hybrid deep learning and ant colony optimization for ocular optical coherence tomography image classification

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

Published: Feb. 18, 2025

Language: Английский

---

CS U‐NET: A Medical Image Segmentation Method Integrating Spatial and Contextual Attention Mechanisms Based on U‐NET

International Journal of Imaging Systems and Technology, Journal Year: 2025, Volume and Issue: 35(2)

Published: March 1, 2025

ABSTRACT Medical image segmentation is a crucial process in medical analysis, with convolutional neural network (CNN)‐based methods achieving notable success recent years. Among these, U‐Net has gained widespread use due to its simple yet effective architecture. However, CNNs still struggle capture global, long‐range semantic information. To address this limitation, we present CS U‐NET, novel method built upon Swin‐U‐Net, which integrates spatial and contextual attention mechanisms. This hybrid approach combines the strengths of both transformers architectures enhance performance. In framework, tokenized patches are processed through transformer‐based U‐shaped encoder‐decoder, enabling learning local global features via skip connections. Our achieves Dice Similarity Coefficient 78.64% 95% Hausdorff distance 21.25 on Synapse multiorgan dataset, outperforming Trans‐U‐Net other state‐of‐the‐art variants by 4% 6%, respectively. The experimental results highlight significant improvements prediction accuracy edge detail preservation provided our approach.

Language: Английский

---

Generalized Framework for Liquid Neural Network upon Sequential and Non-Sequential Tasks

Mathematics, Journal Year: 2024, Volume and Issue: 12(16), P. 2525 - 2525

Published: Aug. 15, 2024

This paper introduces a novel approach to neural networks: Generalized Liquid Neural Network (GLNN) framework. design excels at handling both sequential and non-sequential tasks. By leveraging the Runge Kutta DOPRI method, GLNN enables dynamic simulation of complex systems across diverse fields. Our research demonstrates framework’s capabilities through three key applications. In predicting damped sinusoidal trajectories, LNN outperforms ODE by approximately 46.03% conventional 57.88%. Modelling non-linear RLC circuits shows 20% improvement in precision. Finally, medical diagnosis Optical Coherence Tomography (OCT) image analysis, our achieves an F1 score 0.98, surpassing classical 10%. These advancements signify significant shift, opening new possibilities for networks system modelling healthcare diagnostics. advances field introducing versatile reliable network architecture.

Language: Английский

---

Precision Segmentation of Subretinal Fluids in OCT Using Multiscale Attention-Based U-Net Architecture

Bioengineering, Journal Year: 2024, Volume and Issue: 11(10), P. 1032 - 1032

Published: Oct. 16, 2024

This paper presents a deep-learning architecture for segmenting retinal fluids in patients with Diabetic Macular Oedema (DME) and Age-related Degeneration (AMD). Accurate segmentation of multiple fluid types is critical diagnosis treatment planning, but existing techniques often struggle precision. We propose an encoder–decoder network inspired by U-Net, processing enhanced OCT images their edge maps. The encoder incorporates Residual Inception modules autoencoder-based multiscale attention mechanism to extract detailed features. Our method shows superior performance across several datasets. On the RETOUCH dataset, achieved F1 Scores 0.82 intraretinal (IRF), 0.93 subretinal (SRF), 0.94 pigment epithelial detachment (PED). model also performed well on OPTIMA DUKE datasets, demonstrating high precision, recall, Scores. significantly enhances accuracy offering valuable tool diagnosing managing diseases. Its integration dual-input processing, attention, advanced highlights its potential improve clinical outcomes advance disease treatment.

Language: Английский

---

Deep learning based highly accurate transplanted bioengineered corneal equivalent thickness measurement using optical coherence tomography

npj Digital Medicine, Journal Year: 2024, Volume and Issue: 7(1)

Published: Nov. 5, 2024

Corneal transplantation is the primary treatment for irreversible corneal diseases, but due to limited donor availability, bioengineered equivalents are being developed as a solution, with biocompatibility, structural integrity, and physical function considered key factors. Since conventional evaluation methods may not fully capture complex properties of cornea, there need advanced imaging assessment techniques. In this study, we proposed deep learning-based automatic segmentation method transplanted using optical coherence tomography achieve highly accurate graft integrity biocompatibility. Our provides quantitative individual thickness values, detailed maps, volume measurements equivalents, has been validated through 14 days monitoring. Based on results, it expected have high clinical utility human keratoplasties, including opacity area implanted part extraction, beyond animal studies.

Language: Английский

---

AGSAM: Agent-Guided Segment Anything Model for Automatic Segmentation in Few-Shot Scenarios

Bioengineering, Journal Year: 2024, Volume and Issue: 11(5), P. 447 - 447

Published: April 30, 2024

Precise medical image segmentation of regions interest (ROIs) is crucial for accurate disease diagnosis and progression assessment. However, acquiring high-quality annotated data at the pixel level poses a significant challenge due to resource-intensive nature this process. This scarcity results in few-shot scenarios, which are highly prevalent clinical applications. To address obstacle, paper introduces Agent-Guided SAM (AGSAM), an innovative approach that transforms Segment Anything Model (SAM) into fully automated method by automating prompt generation. Capitalizing on pre-trained feature extraction decoding capabilities SAM-Med2D, AGSAM circumvents need manual engineering, ensuring adaptability across diverse methods. Furthermore, proposed augmentation convolution module (FACM) enhances model accuracy promoting stable representations. Experimental evaluations demonstrate AGSAM’s consistent superiority over other methods various metrics. These findings highlight efficacy tackling challenges associated with limited while achieving segmentation.

Language: Английский

---

Circumpapillary OCT-based multi-sector analysis of retinal layer thickness in patients with glaucoma and high myopia

Computerized Medical Imaging and Graphics, Journal Year: 2024, Volume and Issue: 118, P. 102464 - 102464

Published: Nov. 19, 2024

Language: Английский

---