SNC_Net: Skin Cancer Detection by Integrating Handcrafted and Deep Learning-Based Features Using Dermoscopy Images

Mathematics, Journal Year: 2024, Volume and Issue: 12(7), P. 1030 - 1030

Published: March 29, 2024

The medical sciences are facing a major problem with the auto-detection of disease due to fast growth in population density. Intelligent systems assist professionals early detection and also help provide consistent treatment that reduces mortality rate. Skin cancer is considered be deadliest most severe kind cancer. Medical utilize dermoscopy images make manual diagnosis skin This method labor-intensive time-consuming demands considerable level expertise. Automated methods necessary for occurrence hair air bubbles dermoscopic affects research aims classify eight different types cancer, namely actinic keratosis (AKs), dermatofibroma (DFa), melanoma (MELa), basal cell carcinoma (BCCa), squamous (SCCa), melanocytic nevus (MNi), vascular lesion (VASn), benign (BKs). In this study, we propose SNC_Net, which integrates features derived from through deep learning (DL) models handcrafted (HC) feature extraction aim improving performance classifier. A convolutional neural network (CNN) employed classification. Dermoscopy publicly accessible ISIC 2019 dataset utilized train validate model. proposed model compared four baseline models, EfficientNetB0 (B1), MobileNetV2 (B2), DenseNet-121 (B3), ResNet-101 (B4), six state-of-the-art (SOTA) classifiers. With an accuracy 97.81%, precision 98.31%, recall 97.89%, F1 score 98.10%, outperformed SOTA classifiers as well models. Moreover, Ablation study performed on its performance. therefore assists dermatologists other detection.

Language: Английский

---

Enhancing Skin Cancer Diagnosis Using Swin Transformer with Hybrid Shifted Window-Based Multi-head Self-attention and SwiGLU-Based MLP

Deleted Journal, Journal Year: 2024, Volume and Issue: unknown

Published: June 5, 2024

Abstract Skin cancer is one of the most frequently occurring cancers worldwide, and early detection crucial for effective treatment. Dermatologists often face challenges such as heavy data demands, potential human errors, strict time limits, which can negatively affect diagnostic outcomes. Deep learning–based systems offer quick, accurate testing enhanced research capabilities, providing significant support to dermatologists. In this study, we Swin Transformer architecture by implementing hybrid shifted window-based multi-head self-attention (HSW-MSA) in place conventional (SW-MSA). This adjustment enables model more efficiently process areas skin overlap, capture finer details, manage long-range dependencies, while maintaining memory usage computational efficiency during training. Additionally, study replaces standard multi-layer perceptron (MLP) with a SwiGLU-based MLP, an upgraded version gated linear unit (GLU) module, achieve higher accuracy, faster training speeds, better parameter efficiency. The modified model-base was evaluated using publicly accessible ISIC 2019 dataset eight classes compared against popular convolutional neural networks (CNNs) cutting-edge vision transformer (ViT) models. exhaustive assessment on unseen test dataset, proposed Swin-Base demonstrated exceptional performance, achieving accuracy 89.36%, recall 85.13%, precision 88.22%, F1-score 86.65%, surpassing all previously reported deep learning models documented literature.

Language: Английский

---

A robust deep learning framework for multiclass skin cancer classification

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

Published: Feb. 10, 2025

Skin cancer represents a significant global health concern, where early and precise diagnosis plays pivotal role in improving treatment efficacy patient survival rates. Nonetheless, the inherent visual similarities between benign malignant lesions pose substantial challenges to accurate classification. To overcome these obstacles, this study proposes an innovative hybrid deep learning model that combines ConvNeXtV2 blocks separable self-attention mechanisms, tailored enhance feature extraction optimize classification performance. The inclusion of initial two stages is driven by their ability effectively capture fine-grained local features subtle patterns, which are critical for distinguishing visually similar lesion types. Meanwhile, adoption later allows selectively prioritize diagnostically relevant regions while minimizing computational complexity, addressing inefficiencies often associated with traditional mechanisms. was comprehensively trained validated on ISIC 2019 dataset, includes eight distinct skin categories. Advanced methodologies such as data augmentation transfer were employed further robustness reliability. proposed architecture achieved exceptional performance metrics, 93.48% accuracy, 93.24% precision, 90.70% recall, 91.82% F1-score, outperforming over ten Convolutional Neural Network (CNN) based Vision Transformer (ViT) models tested under comparable conditions. Despite its robust performance, maintains compact design only 21.92 million parameters, making it highly efficient suitable deployment. Proposed Model demonstrates accuracy generalizability across diverse classes, establishing reliable framework clinical practice.

Language: Английский

---

Diagnosis and prognosis of melanoma from dermoscopy images using machine learning and deep learning: a systematic literature review

BMC Cancer, Journal Year: 2025, Volume and Issue: 25(1)

Published: Jan. 13, 2025

Melanoma is a highly aggressive skin cancer, where early and accurate diagnosis crucial to improve patient outcomes. Dermoscopy, non-invasive imaging technique, aids in melanoma detection but can be limited by subjective interpretation. Recently, machine learning deep techniques have shown promise enhancing diagnostic precision automating the analysis of dermoscopy images. This systematic review examines recent advancements (ML) (DL) applications for prognosis using We conducted thorough search across multiple databases, ultimately reviewing 34 studies published between 2016 2024. The covers range model architectures, including DenseNet ResNet, discusses datasets, methodologies, evaluation metrics used validate performance. Our results highlight that certain such as DCNN demonstrated outstanding performance, achieving over 95% accuracy on HAM10000, ISIC other datasets from provides insights into strengths, limitations, future research directions methods prognosis. It emphasizes challenges related data diversity, interpretability, computational resource requirements. underscores potential transform through improved efficiency. Future should focus creating accessible, large interpretability increase clinical applicability. By addressing these areas, models could play central role advancing care.

Language: Английский

---

Towards unbiased skin cancer classification using deep feature fusion

BMC Medical Informatics and Decision Making, Journal Year: 2025, Volume and Issue: 25(1)

Published: Jan. 31, 2025

Abstract This paper introduces SkinWiseNet (SWNet), a deep convolutional neural network designed for the detection and automatic classification of potentially malignant skin cancer conditions. SWNet optimizes feature extraction through multiple pathways, emphasizing width augmentation to enhance efficiency. The proposed model addresses potential biases associated with conditions, particularly in individuals darker tones or excessive hair, by incorporating fusion assimilate insights from diverse datasets. Extensive experiments were conducted using publicly accessible datasets evaluate SWNet’s effectiveness.This study utilized four datasets-Mnist-HAM10000, ISIC2019, ISIC2020, Melanoma Skin Cancer-comprising images categorized into benign classes. Explainable Artificial Intelligence (XAI) techniques, specifically Grad-CAM, employed interpretability model’s decisions. Comparative analysis was performed three pre-existing learning networks-EfficientNet, MobileNet, Darknet. results demonstrate superiority, achieving an accuracy 99.86% F1 score 99.95%, underscoring its efficacy gradient propagation capture across various levels. research highlights significant advancing classification, providing robust tool accurate early diagnosis. integration enhances mitigates hair tones. outcomes this contribute improved patient healthcare practices, showcasing exceptional capabilities classification.

Language: Английский

---

A fuzzy rank-based deep ensemble methodology for multi-class skin cancer classification

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

Published: Feb. 20, 2025

Skin cancer is widespread and can be potentially fatal. According to the World Health Organisation (WHO), it has been identified as a leading cause of mortality. It essential detect skin early so that effective treatment provided at an initial stage. In this study, widely-used HAM10000 dataset, containing high-resolution images various lesions, employed train evaluate. Our methodology for dataset involves balancing imbalanced by augmenting followed splitting into train, test validation set, preprocessing images, training individual models Xception, InceptionResNetV2 MobileNetV2, then combining their outputs using fuzzy logic generate final prediction. We examined performance ensemble standard metrics like classification accuracy, confusion matrix, etc. achieved impressive accuracy 95.14% result demonstrates effectiveness our approach in accurately identifying lesions. To further assess efficiency model, additional tests have performed on DermaMNIST from MedMNISTv2 collection. The model performs well transcends benchmark 76.8%, achieving 78.25%. Thus efficient classification, showcasing its potential clinical applications.

Language: Английский

---

A Computing Framework for Transfer Learning and Ensemble Classification of Surface Patterns

Journal of Machine and Computing, Journal Year: 2025, Volume and Issue: unknown, P. 140 - 153

Published: Jan. 3, 2025

The rapid increase in population density has posed significant challenges to medical sciences the auto-detection of various diseases. Intelligent systems play a crucial role assisting professionals with early disease detection and providing consistent treatment, ultimately reducing mortality rates. Skin-related diseases, particularly those that can become severe if not detected early, require timely identification expedite diagnosis improve patient outcomes. This paper proposes transfer learning-based ensemble deep learning model for diagnosing dermatological conditions at an stage. Data augmentation techniques were employed number samples create diverse data pattern within dataset. study applied ResNet50, InceptionV3, DenseNet121 models, leading development weighted average model. system was trained tested using International Skin Imaging Collaboration (ISIC) proposed demonstrated superior performance, achieving 98.5% accuracy, 97.50% Kappa, 97.67% MCC (Matthews Correlation Coefficient), 98.50% F1 score. outperformed existing state-of-the-art models classification provides valuable support dermatologists specialists detection. Compared previous research, offers high accuracy lower computational complexity, addressing challenge skin-related

Language: Английский

---

Evaluation of an acne lesion detection and severity grading model for Chinese population in online and offline healthcare scenarios

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

Published: Jan. 7, 2025

Accurate acne severity grading is crucial for effective clinical treatment and timely follow-up management. Although some artificial intelligence methods have been developed to automate the process of grading, diversity image capture sources various application scenarios can affect their performance. Therefore, it's necessary design special evaluate them systematically before introducing into practice. To develop a deep learning-based algorithm that could accurately accomplish lesion detection simultaneously in different healthcare scenarios. We collected 2,157 facial images from two public three self-built datasets model development evaluation. An called AcneDGNet was constructed with feature extraction module, module module. Its performance evaluated both online offline Experimental results on largest most diverse evaluation revealed overall achieved accuracies 89.5% 89.8% For visits scenarios, accuracy detecting changing trends reached 87.8%, total counting error 1.91 ± 3.28 all lesions. Additionally, prospective demonstrated not only much more accurate than junior dermatologists but also comparable senior dermatologists. These findings indicated effectively assist patients diagnosis management acne,

Language: Английский

---

Minimal sourced and lightweight federated transfer learning models for skin cancer detection

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

Published: Jan. 21, 2025

Language: Английский

---

Cilt Kanseri Tanısı için Farklı Evrişimsel Sinir Ağı Modellerinin Karşılaştırılması

Iğdır Üniversitesi Fen Bilimleri Enstitüsü Dergisi, Journal Year: 2025, Volume and Issue: 15(1), P. 25 - 38

Published: Feb. 19, 2025

Son yıllarda, dünya genelinde cilt kanseri görülme oranında önemli bir artış gözlemlenmektedir. Cilt kanserinin zamanında ve doğru şekilde teşhis edilmesi, tedavi başarı oranlarını artırmakta aynı zamanda hastaların yaşam kalitesinin iyileşmesine büyük katkı sağlamaktadır. Geleneksel tanı yöntemleri genellikle görsel değerlendirmelere dayanmakta öznel yaklaşım içermektedir. Bununla birlikte, derin öğrenme algoritmaları, teşhislerinin doğruluğunu verimliliğini artırmak için etkili çözümler sunmaktadır. Bu çalışmada, EfficientNet, VGG, Inception, DenseNet DarkNet gibi gelişmiş Evrişimsel Sinir Ağı (CNN) modellerinin sınıflandırmasındaki performansları incelenmiştir. Toplamda yirmi CNN modeli, ISIC 2017 veri seti üzerinde, artırma transfer teknikleri kullanılarak eğitilmiş detaylı değerlendirilmiştir. Deneysel sonuçlar, EfficientNet-b0 modelinin %84.00 doğruluk, %83.63 kesinlik, %74.96 duyarlılık %78.59 F1-skoru ile en yüksek performansı sergilediğini göstermiştir. kapsamlı analiz, tabanlı modellerin teşhisindeki etkinliğini göstermekte gelecekteki araştırmalar bu algoritmaların potansiyelini ortaya koymaktadır.

---