A Multiscale Wavelet-CNN Fusion Framework for Automated Rheumatoid Arthritis Severity Classification
Engineering, Technology & Applied Science Research · 11.08.2026 · 16 (4) · 37214-37219
DOI: 10.48084/etasr.19465 ↗Kurz gefasst
Fachartikel in der Zeitschrift Engineering, Technology & Applied Science Research, online erschienen am 11.08.2026. Verfasst von Lassaad Ben Ammar. Die Arbeit nennt 16 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Early detection of Rheumatoid Arthritis (RA) is significant to preventing irreversible joint damage. Traditional Convolutional Neural Networks (CNNs) often fail to capture both global structures and fine-grained markers such as bone erosion. This study proposes a novel Dual-Stream Wavelet-Fusion Network (DS-WFN) that incorporates an Adaptive Wavelet-Selection Mechanism, which dynamically selects optimal mother wavelets to isolate specific RA symptoms. A structural stream processes low-frequency coefficients to provide the morphological background, while a parallel detail stream focuses on high-frequency sub-bands to pick up on small changes in texture. An Adaptive Spatial Alignment layer ensures that features from different streams can be combined without any problems. Experimental results using a dataset of 1,485 radiographs demonstrate that the proposed DS-WFN architecture achieves an accuracy of 74%, significantly outperforming standard SOTA and single-stream wavelet-CNN models. This frequency-adaptive design provides a more robust and interpretable tool for clinical RA assessment.
Zitieren
APA
Ammar, L. B. (2026). A Multiscale Wavelet-CNN Fusion Framework for Automated Rheumatoid Arthritis Severity Classification. Engineering, Technology & Applied Science Research, 16(4), 37214-37219. https://doi.org/10.48084/etasr.19465
Vancouver
Ammar LB. A Multiscale Wavelet-CNN Fusion Framework for Automated Rheumatoid Arthritis Severity Classification. Engineering, Technology & Applied Science Research. 2026;16(4):37214-37219. doi:10.48084/etasr.19465
BibTeX
@article{Ammar2026,
author = {Ammar, Lassaad Ben},
title = {A Multiscale Wavelet-CNN Fusion Framework for Automated Rheumatoid Arthritis Severity Classification},
journal = {Engineering, Technology & Applied Science Research},
year = {2026},
volume = {16},
number = {4},
pages = {37214--37219},
doi = {10.48084/etasr.19465}
}
EUVIMED