Multiple Differential Convolution and Local-Variation Attention UNet: Nucleus Semantic Segmentation Based on Multiple Differential Convolution and Local-Variation Attention

Electronics, Год журнала: 2025, Номер 14(6), С. 1058 - 1058

Опубликована: Март 7, 2025

Nucleus accurate segmentation is a crucial task in biomedical image analysis. While convolutional neural networks (CNNs) have achieved notable progress this field, challenges remain due to the complexity and heterogeneity of cell images, especially overlapping regions nuclei. To address limitations current methods, we propose mechanism multiple differential convolution local-variation attention CNNs, leading so-called U-Net (MDLA-UNet). The employs operators capture gradient direction information, improving network’s capability detect edges. utilizes Haar discrete wavelet transforms for level-1 decomposition obtain approximate features, then derives high-frequency features enhance global context local detail variation feature maps. results on MoNuSeg, TNBC, CryoNuSeg datasets demonstrated superior performance proposed method cells having complex boundaries details with respect existing methods. MDLA-UNet presents ability capturing fine edges maps thus improves nuclei blurred regions.

Язык: Английский

---

Deep learning-based image analysis in muscle histopathology using photo-realistic synthetic data

Communications Medicine, Год журнала: 2025, Номер 5(1)

Опубликована: Март 6, 2025

Artificial intelligence (AI), specifically Deep learning (DL), has revolutionized biomedical image analysis, but its efficacy is limited by the need for representative, high-quality large datasets with manual annotations. While latest research on synthetic data using AI-based generative models shown promising results to tackle this problem, several challenges such as lack of interpretability and vast amounts real remain. This study aims introduce a new approach-SYNTA-for generation photo-realistic address associated state-of-the art DL-based analysis. The SYNTA method employs fully parametric approach create training tailored specific tasks. Its applicability tested in context muscle histopathology skeletal evaluated two real-world validate solve complex analysis tasks data. Here we show that enables expert-level segmentation unseen only By addressing representative data, achieves robust performance offering scalable, controllable interpretable alternative Generative Adversarial Networks (GANs) or Diffusion Models. demonstrates great potential accelerate improve ability generate reduces reliance extensive collection annotations, paving way advancements medical research.

Язык: Английский

---

Lightweight Evolving U-Net for Next-Generation Biomedical Imaging

Diagnostics, Год журнала: 2025, Номер 15(9), С. 1120 - 1120

Опубликована: Апрель 28, 2025

Background/Objectives: Accurate and efficient segmentation of cell nuclei in biomedical images is critical for a wide range clinical research applications, including cancer diagnostics, histopathological analysis, therapeutic monitoring. Although U-Net its variants have achieved notable success medical image segmentation, challenges persist balancing accuracy with computational efficiency, especially when dealing large-scale datasets resource-limited settings. This study aims to develop lightweight scalable U-Net-based architecture that enhances performance while substantially reducing overhead. Methods: We propose novel evolving integrates multi-scale feature extraction, depthwise separable convolutions, residual connections, attention mechanisms improve robustness across diverse imaging conditions. Additionally, we incorporate channel reduction expansion strategies inspired by ShuffleNet minimize model parameters without sacrificing precision. The was extensively validated using the 2018 Data Science Bowl dataset. Results: Experimental evaluation demonstrates proposed achieves Dice Similarity Coefficient (DSC) 0.95 an 0.94, surpassing state-of-the-art benchmarks. effectively delineates complex overlapping structures high fidelity, maintaining efficiency suitable real-time applications. Conclusions: variant offers adaptable solution tasks. Its strong both highlights potential deployment diagnostics biological research, paving way resource-conscious solutions.

Язык: Английский

---

NISNet3D: three-dimensional nuclear synthesis and instance segmentation for fluorescence microscopy images

Scientific Reports, Год журнала: 2023, Номер 13(1)

Опубликована: Июнь 12, 2023

The primary step in tissue cytometry is the automated distinction of individual cells (segmentation). Since cell borders are seldom labeled, generally segmented by their nuclei. While tools have been developed for segmenting nuclei two dimensions, segmentation three-dimensional volumes remains a challenging task. lack effective methods represents bottleneck realization potential cytometry, particularly as clearing present opportunity to characterize entire organs. Methods based on deep learning shown enormous promise, but implementation hampered need large amounts manually annotated training data. In this paper, we describe 3D Nuclei Instance Segmentation Network (NISNet3D) that directly segments through use modified U-Net, marker-controlled watershed transform, and instance system separating touching NISNet3D unique it provides accurate even image using network trained synthetic derived from relatively few volumes, or data obtained without volumes. We quantitative comparison results with variety existing techniques. also examine performance when no ground truth available only were used training.

Язык: Английский

---

DaCSeg: Divide and conquer for accurate overlapping chromosome instance segmentation in metaphase cell images

Biomedical Signal Processing and Control, Год журнала: 2024, Номер 89, С. 105880 - 105880

Опубликована: Янв. 3, 2024

Язык: Английский

---

Automated 3D cytoplasm segmentation in soft X-ray tomography

iScience, Год журнала: 2024, Номер 27(6), С. 109856 - 109856

Опубликована: Апрель 29, 2024

Cells' structure is key to understanding cellular function, diagnostics, and therapy development. Soft X-ray tomography (SXT) a unique tool image without fixation or labeling at high spatial resolution throughput. Fast acquisition times increase demand for accelerated analysis, like segmentation. Currently, segmenting structures done manually major bottleneck in the SXT data analysis. This paper introduces ACSeg, an automated 3D cytoplasm segmentation model. ACSeg generated using semi-automated labels U-Net trained on 43 tomograms of immune T cells, rapidly converging high-accuracy segmentation, therefore reducing time labor. Furthermore, adding only 6 other cell types diversifies model, showing potential optimal experimental design. successfully segmented unseen published Biomedisa, enabling high-throughput analysis volume diverse types.

Язык: Английский

---

Improving generalization capability of deep learning-based nuclei instance segmentation by non-deterministic train time and deterministic test time stain normalization

Computational and Structural Biotechnology Journal, Год журнала: 2024, Номер 23, С. 669 - 678

Опубликована: Янв. 4, 2024

Язык: Английский

---

Comprehensive data analysis of white blood cells with classification and segmentation by using deep learning approaches

Cytometry Part A, Год журнала: 2024, Номер 105(7), С. 501 - 520

Опубликована: Апрель 2, 2024

Deep learning approaches have frequently been used in the classification and segmentation of human peripheral blood cells. The common feature previous studies was that they more than one dataset, but them separately. No study has found combines two datasets to use together. In classification, five types white cells were identified by using a mixture four different datasets. segmentation, determined, three neural networks, including CNN (Convolutional Neural Network), UNet SegNet, applied. results presented compared with those related studies. balanced accuracy 98.03%, test train-independent dataset determined be 97.27%. For rates 98.9% for train-dependent 92.82% proposed obtained both nucleus cytoplasm detection. study, method showed it could detect from high accuracy. Additionally, is promising as diagnostic tool can clinical field, successful segmentation.

Язык: Английский

---

An optimized ensemble model bfased on cuckoo search with Levy Flight for automated gastrointestinal disease detection

Multimedia Tools and Applications, Год журнала: 2024, Номер unknown

Опубликована: Апрель 2, 2024

Язык: Английский

---

Review of cervical cell segmentation

Multimedia Tools and Applications, Год журнала: 2024, Номер unknown

Опубликована: Авг. 2, 2024

Язык: Английский

---