Development of Hybrid Deep Learning Systems Combining Convolutional Neural Networks and Vision Transformers for Medical Ultrasound Image Classification
Diagnostics · 04.11.2026 · 16 (21) · 3640
DOI: 10.3390/diagnostics16213640 ↗Kurz gefasst
Fachartikel in der Zeitschrift Diagnostics, online erschienen am 04.11.2026. Verfasst von Hasan Ucuzal und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Biostatistics and Medical Informatics, Faculty of Medicine, Inonu University, 44280 Malatya, Turkey. Die Arbeit nennt 47 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Inonu University, Inonu University Scientific Research Projects Coordination Unit.
Abstract (Originalsprache)
Background
Accurate classification of thyroid nodule ultrasound images is clinically significant. This study evaluated whether hybrid architectures combining the local feature extraction of convolutional neural networks (CNNs) with the global context modelling of vision transformers (ViTs) outperform each category in isolation.
Methods
Two public datasets (6905 images; 3331 benign, 3574 malignant) were used. Eleven ImageNet-pretrained architectures were compared: CNN-based (ResNet-50, EfficientNet-B4, ConvNeXt), ViT-based (Swin Transformer, DeiT, PVTv2, XCiT) and hybrid (ConViT, CoaT, MobileViT, HuviT). Backbones were frozen and only the classification head was trained; HuviT’s new projection and encoder remained frozen at random initialisation. Wilson and Newcombe 95% confidence intervals (CIs) were computed; explainability used gradient-weighted class activation mapping (Grad-CAM).
Results
ConViT achieved the highest accuracy (83.30%; 95% CI 80.91–85.45) and an area under the receiver operating characteristic curve (ROC-AUC) of 89.94% (95% CI 87.8–91.7). ConvNeXt ranked second (82.63%) but had the highest ROC-AUC (90.02%; 95% CI 87.9–91.8); the intervals overlapped. HuviT reached 76.35%. Grad-CAM for a representative case highlighted lesion areas.
Conclusions
While ConViT achieved the highest performance on threshold-based metrics, these findings suggest the potential of specific hybrid designs rather than establishing general superiority of hybrid architectures. Given the retrospective design and the partial association between image source and class label, prospective external validation is warranted.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Biostatistics and Medical Informatics, Faculty of Medicine, Inonu University, 44280 Malatya, Turkey.
ORCID-Profile: Hasan Ucuzal, Cemil Colak, Emek Guldogan
Förderung
- Inonu University
- Inonu University Scientific Research Projects Coordination Unit
Zitieren
APA
Ucuzal, H., Colak, C., & Guldogan, E. (2026). Development of Hybrid Deep Learning Systems Combining Convolutional Neural Networks and Vision Transformers for Medical Ultrasound Image Classification. Diagnostics, 16(21), 3640. https://doi.org/10.3390/diagnostics16213640
Vancouver
Ucuzal H, Colak C, Guldogan E. Development of Hybrid Deep Learning Systems Combining Convolutional Neural Networks and Vision Transformers for Medical Ultrasound Image Classification. Diagnostics. 2026;16(21):3640. doi:10.3390/diagnostics16213640
BibTeX
@article{Ucuzal2026,
author = {Ucuzal, Hasan and Colak, Cemil and Guldogan, Emek},
title = {Development of Hybrid Deep Learning Systems Combining Convolutional Neural Networks and Vision Transformers for Medical Ultrasound Image Classification},
journal = {Diagnostics},
year = {2026},
volume = {16},
number = {21},
pages = {3640},
doi = {10.3390/diagnostics16213640}
}
EUVIMED