Multi-Skin disease classification using hybrid deep learning model

Technology and Health Care, Год журнала: 2025, Номер unknown

Опубликована: Фев. 2, 2025

Among the many cancers that people face today, skin cancer is among deadliest and most dangerous. As a result, improving patients’ chances of survival requires to be identified classified early. Therefore, it critical assist radiologists in detecting through development Computer Aided Diagnosis (CAD) techniques. The diagnostic procedure currently makes heavy use Deep Learning (DL) techniques for disease identification. In addition, lesion extraction improved classification performance are achieved Region Growing (RG) based segmentation. At outset this study, noise reduced using an Adaptive Wiener Filter (AWF), hair removed Maximum Gradient Intensity (MGI). Then, best RG, which result integrating RG with Modified Honey Badger Optimiser (MHBO), does Finally, several forms DL model MobileSkinNetV2. experiments were conducted on ISIC dataset results show accuracy precision 99.01% 98.6%, respectively. comparison existing models, experimental proposed performs competitively, great news dermatologists treating cancer.

Язык: Английский

---

Dermo‐Optimizer: Skin Lesion Classification Using Information‐Theoretic Deep Feature Fusion and Entropy‐Controlled Binary Bat Optimization

International Journal of Imaging Systems and Technology, Год журнала: 2024, Номер 34(5)

Опубликована: Сен. 1, 2024

ABSTRACT Increases in the prevalence of melanoma, most lethal form skin cancer, have been observed over last few decades. However, likelihood a longer life span for individuals is considerably improved with early detection this malignant illness. Even though field computer vision has attained certain level success, there still degree ambiguity that represents an unresolved research challenge. In initial phase study, primary objective to improve information derived from input features by combining multiple deep models proposed Information‐theoretic feature fusion method. Subsequently, second phase, study aims decrease redundant and noisy through down‐sampling using entropy‐controlled binary bat selection algorithm. The methodology effectively maintains integrity original space, resulting creation highly distinctive information. order obtain desired set features, three contemporary are employed via transfer learning: Inception‐Resnet V2, DenseNet‐201, Nasnet Mobile. By techniques, we may fuse significant amount into vector subsequently remove any effectiveness supported evaluation conducted on well‐known dermoscopic datasets, specifically , ISIC‐2016, ISIC‐2017. validate approach, several performance indicators taken account, such as accuracy, sensitivity, specificity, false negative rate (FNR), positive (FPR), F1‐score. accuracies obtained all datasets utilizing 99.05%, 96.26%, 95.71%, respectively.

Язык: Английский

---

Deep Learning Model for Metal Gear Defect Detection

Lecture notes in electrical engineering, Год журнала: 2025, Номер unknown, С. 132 - 136

Опубликована: Янв. 1, 2025

Язык: Английский

---

Skin lesion segmentation with a multiscale input fusion U-Net incorporating Res2-SE and pyramid dilated convolution

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Март 7, 2025

Skin lesion segmentation is crucial for identifying and diagnosing skin diseases. Accurate aids in localizing diseases, monitoring morphological changes, extracting features further diagnosis, especially the early detection of cancer. This task challenging due to irregularity lesions dermatoscopic images, significant color variations, boundary blurring, other complexities. Artifacts like hairs, blood vessels, air bubbles complicate automatic segmentation. Inspired by U-Net its variants, this paper proposes a Multiscale Input Fusion Residual Attention Pyramid Convolution Network (MRP-UNet) dermoscopic image MRP-UNet includes three modules: Module (MIF), Res2-SE Module, Dilated (PDC). The MIF module processes different sizes morphologies fusing input information from various scales. integrates Res2Net SE mechanisms enhance multi-scale feature extraction. PDC captures at receptive fields through pyramid dilated convolution, improving accuracy. Experiments on ISIC 2016, 2017, 2018, PH2, HAM10000 datasets show that outperforms methods. Ablation studies confirm effectiveness main modules. Both quantitative qualitative analyses demonstrate MRP-UNet's superiority over state-of-the-art enhances combining multiscale fusion, residual attention, convolution. It achieves higher accuracy across multiple datasets, showing promise disease diagnosis improved patient outcomes.

Язык: Английский

---

A Review of U‐Net‐Based Deep Learning Models for Skin Lesion Segmentation

International Journal of Imaging Systems and Technology, Год журнала: 2025, Номер 35(3)

Опубликована: Май 1, 2025

ABSTRACT Automated skin lesion segmentation is crucial for early and accurate cancer diagnosis. Deep learning, particularly U‐Net, has revolutionized the field of automatic segmentation. This review comprehensively examines U‐Net its variants employed automated It outlines foundational architecture explores diverse architectural innovations, including attention mechanisms, advanced skip connections, residual dilated convolutions, transformer models, hybrid models. The highlights how these adaptations address inherent challenges in segmentation, data limitations heterogeneity. also discusses commonly used datasets, evaluation metrics, compares model performance computational cost. Finally, it addresses existing future research directions to advance

Язык: Английский

---

Hybrid MultiResUNet with transformers for medical image segmentation

Biomedical Signal Processing and Control, Год журнала: 2025, Номер 110, С. 108056 - 108056

Опубликована: Май 23, 2025

Язык: Английский

---

Wavelet Guided Visual State Space Model and Patch Resampling Enhanced U-shaped Structure for Skin Lesion Segmentation

IEEE Access, Год журнала: 2024, Номер 12, С. 181521 - 181532

Опубликована: Янв. 1, 2024

Язык: Английский

---

MSAByNet: A multiscale subtraction attention network framework based on Bayesian loss for medical image segmentation

Biomedical Signal Processing and Control, Год журнала: 2024, Номер 103, С. 107393 - 107393

Опубликована: Дек. 24, 2024

Язык: Английский

---

Performance Evaluation of U-Net Based Methods for Lesion Segmentation from Dermoscopy Images

Proceedings of international conference on intelligent systems and new applications., Год журнала: 2024, Номер unknown

Опубликована: Апрель 28, 2024

Skin cancer accounts for approximately half of all cases worldwide, making it one the most prevalent types cancer. Melanoma, which develops from melanocytes that give skin its color, is lethal among cancers. Early diagnosis melanoma, particularly through dermoscopy images, vital importance. To this end, automated diagnostic systems significantly aid dermatologists in their decision-making processes. In recent years, advancements deep learning and machine have improved accuracy. Specifically, CNN-based algorithms are utilized medical image analysis lesion segmentation. While traditional methods struggle to capture fine details broader context, U-Net architecture overcomes these challenges, providing more accurate This study evaluates U-Net, Residual Attention models The performance measured using Dice Score, Jaccard Index, train loss metrics. results reveal demonstrates highest performance, with a Score 0.8063 Index 0.7203.

Язык: Английский

---

Deep-Multiscale Stratified Aggregation

Research Square (Research Square), Год журнала: 2024, Номер unknown

Опубликована: Ноя. 5, 2024

In deep learning based vision tasks, improving multiscale representation by combining shallow and features has consistently led to performance gains across a wide range of applications. However, significant discrepancies in both scale semantic content often occur during the fusion features. Most existing approaches rely on standard convolutional structures for representing features, which may not fully capture complexity underlying data. To address this, we propose novel deep-multiscale stratified aggregation (D-MSA) module, could improve extraction efficiently aggregating multiple receptive fields. The D-MSA module was integrated into YOLO architecture enhance capacity processing complex Experiments PASCAL VOC 2012 dataset demonstrate that effectively handle while computational efficiency, making it suitable object detection challenging environments.

Язык: Английский

---