Skip to main content
2025

From survey to solution: A deep learning framework for reliable monkeypox diagnosis using skin images

Md Shakhawat Hossain, Munim Ahmed, Md Sahilur Rahman

Array , Vol. 28

From survey to solution: A deep learning framework for reliable monkeypox diagnosis using skin images

Abstract

Monkeypox, a re-emerging zoonotic disease, poses a global health threat due to its rapid transmission and visual similarity to other skin lesions such as chickenpox, measles and acne. Deep learning methods, which detect monkeypox from skin images, offer a promising solution to overcome the limitations of manual and PCR-based diagnoses, which are time-consuming, error-prone and impractical in low-resource settings. However, existing methods are limited by poor dataset quality, weak generalizability and inconsistent benchmarking. Practical issues, such as lesion variability, image noise, unsuitable augmentations and minimal preprocessing, pose further challenges to clinical deployment. This study addressed these issues through a three-fold contribution: a comprehensive survey of deep learning methods analyzing their strengths and limitations; the development of a diverse, clinically representative benchmark dataset to better assess model generalizability; and a robust deep learning ensemble framework that improved diagnostic accuracy across diverse skin images. The proposed ensemble method incorporated practical classes, a noise-free and balanced dataset, clinically relevant augmentations and effective preprocessing steps, achieving over 95% accuracy across major public datasets to ensure robustness and readiness for clinical deployment. Explainability analysis using Shapley Additive exPlanation (SHAP) confirmed the method’s reliability across all skin tones and body parts. A paired t-test showed that the ensemble model performed significantly better than individual models across four public datasets (𝑝 = 0.005, Cohen’s 𝑑 = 2.38).

Citation

Md Shakhawat Hossain, Munim Ahmed, Md Sahilur Rahman. "From survey to solution: A deep learning framework for reliable monkeypox diagnosis using skin images." Array 28 (2025).

BibTeX

@article{pub48_2025,
  title={From survey to solution: A deep learning framework for reliable monkeypox diagnosis using skin images},
  author={Md Shakhawat Hossain, Munim Ahmed, Md Sahilur Rahman},
  journal={Array},
  volume={28},
  year={2025},
  doi={https://doi.org/10.1016/j.array.2025.100554}
}
Publication Details
Type:
Year:
2025
Journal:
Array
Volume:
28
Share

Related Publications

SwiftMSeg: lightweight multi-scale local–global context modeling with transformer for medical image segmentation
2026
SwiftMSeg: lightweight multi-scale local–global context mod…

Jahid Hasan Rony, Md Shakhawat Hossain & Fazlul Hasan Siddiqui

LGGC-Net: a local-global graph and color attention-based lightweight CNN for skin cancer classification
2026
LGGC-Net: a local-global graph and color attention-based li…

Md Aminur Sarker, Md Alamgir Kabir, Md Shakhawat Hossain

SiNuS: A Comprehensive Dataset for Singular Nuclei Segmentation for HER2 Grading of Breast Cancer
2026
SiNuS: A Comprehensive Dataset for Singular Nuclei Segmenta…

Md Shakhawat Hossain, Md Sahilur Rahman, Munim Ahmed