Multi-Classification of Skin Lesion Images Including Mpox Disease Using Transformer-Based Deep Learning Architectures

Diagnostics, Год журнала: 2025, Номер 15(3), С. 374 - 374

Опубликована: Фев. 5, 2025

Background/Objectives: As reported by the World Health Organization, Mpox (monkeypox) is an important disease present in 110 countries, mostly South Asia and Africa. The number of cases has increased rapidly, medical world worried about emergence a new pandemic. Detection traditional methods (using test kits) costly slow process. For this reason, there need for that have high success rates can diagnose from skin images with deep-learning-based autonomous method. Methods: In work, we propose multi-class, fast, reliable diagnosis model using transformer-based deep learning architectures lesion images, including disease. Our other aim to investigate effects self-supervised learning, self-distillation, shifted window techniques on classification when multi-class are trained architectures. Skin Lesion Dataset, Version 2.0, which was publicly released 2024, used training, validation, testing processes study. Results: SwinTransformer architecture proposed our study achieved 8% higher accuracy evaluation metric compared its closest competitor literature. ViT, MAE, DINO, 93.10%, 84.60%, 90.40%, 93.71% success, respectively. Conclusions: results obtained showed be diagnosed support doctors decision-making. addition, provides fields where low terms technique use.

Язык: Английский

Explainable AI for Symptom-Based Detection of Monkeypox: a machine learning approach

BMC Infectious Diseases, Год журнала: 2025, Номер 25(1)

Опубликована: Март 26, 2025

Monkeypox, a viral zoonotic disease, is an emerging global health concern, with rising incidence and outbreaks extending beyond its endemic regions in Central and, West Africa the world. The disease transmits through contact infected animals humans, leading to fever, rash, lymphadenopathy symptoms. Control efforts include surveillance, tracing, vaccination campaigns; however, increasing number of cases underscores necessity for coordinated response mitigate impact. Since monkeypox has become public issue, new methods efficiently identifying are required. control infections depends on early detection prediction. This study aimed utilize Symptom-Based Detection Monkeypox using machine-learning approach. research presents machine learning approach that integrates various Explainable Artificial Intelligence (XAI) enhance based clinical symptoms, addressing limitations image-based diagnostic systems. In this study, we used publicly available dataset from GitHub containing features about disease. data have been analysed Random Forest, Bagging, Gradient Boosting, CatBoost, XGBoost, LGBMClassifier develop robust predictive model. shows models can accurately diagnose symptoms like other By XAI techniques feature importance, not only achieved high accuracy but also provided transparency decision-making. integration explainable intelligence (AI) enhances trust allows healthcare professionals understand predictions, timely interventions improved responses outbreaks. All Machine compared evaluation matrix. best performance was LGBMClassifier, 89.3%. addition, multiple Techniques tools were help examining explaining output Our combining AI greatly case boosts medical professionals. These result directly involving reader care professional decision-making process, making informed decisions, allocating resources by providing insight into process. potential particularly enhancing infectious diseases such as monkeypox.

Язык: Английский