EUVIMEDEUVIMED entdecken
Crossref

ICARECS-2026 - Hybrid Cnn–vision Transformer Framework for Automated Medical Image Classification

A.Philomina Jenifer

i-manager's Journal on Pattern Recognition · 20.06.2026 · 13 (1) · 1

DOI: 10.26634/jpr.13.1.11343 ↗

Kurz gefasst

Fachartikel in der Zeitschrift i-manager's Journal on Pattern Recognition, online erschienen am 15.06.2026. Verfasst von A.Philomina Jenifer.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Correct and automatic diagnosis of medical images is extremely important for early diagnosis and clinical decision support. ConvNets have been shown to be good at learning local spatial properties like edges and textures, but do not easily learn long-range dependencies in the image. In contrast, the Vision Transformers (ViTs) learn local contextual information via self-attention, but generally need to have a large training set and a weak inductive bias. To address the limitations of these approaches, in this paper, we present a hybrid network of CNN–Vision Transformer by leveraging the complementary advantages of both architectures for medical image classification. CNN features rich local feature maps are extracted, tokenized and fed to a Transformer encoder to capture global dependencies between features. A lightweight fully connected head classifies the fused representation. Evaluation results were obtained on a publicly available medical imaging dataset and showed that the proposed framework achieved an accuracy of 97.2%, which is higher than the standalone CNN (94.1%) and ViT (93.5%) baselines. The results show that hybrid structures have a strong potential for reliable and data-efficient medical image analysis. Keywords: Medical image classification, Convolutional Neural Network, Vision Transformer, Hybrid deep learning, Computer-aided diagnosis ________________________________________

Zitieren

APA

Jenifer, A. (2026). ICARECS-2026 - Hybrid Cnn–vision Transformer Framework for Automated Medical Image Classification. i-manager's Journal on Pattern Recognition, 13(1), 1. https://doi.org/10.26634/jpr.13.1.11343

Vancouver

Jenifer A. ICARECS-2026 - Hybrid Cnn–vision Transformer Framework for Automated Medical Image Classification. i-manager's Journal on Pattern Recognition. 2026;13(1):1. doi:10.26634/jpr.13.1.11343

BibTeX
@article{Jenifer2026,
  author = {Jenifer, A.Philomina},
  title = {ICARECS-2026 - Hybrid Cnn–vision Transformer Framework for Automated Medical Image Classification},
  journal = {i-manager's Journal on Pattern Recognition},
  year = {2026},
  volume = {13},
  number = {1},
  pages = {1},
  doi = {10.26634/jpr.13.1.11343}
}
Kostenlos starten →