TY  - JOUR
AU  - Khan, Ikram Majeed
AU  - Khan, Wisal
PY  - 2026
DA  - 2026/06/28
TI  - Scale-Specific Visual Sensing for Colonoscopy Polyp Segmentation via Hybrid CNN-Transformer Attention
JO  - ICCK Transactions on Sensing, Communication, and Control
T2  - ICCK Transactions on Sensing, Communication, and Control
JF  - ICCK Transactions on Sensing, Communication, and Control
VL  - 3
IS  - 2
SP  - 109
EP  - 123
DO  - 10.62762/TSCC.2026.664028
UR  - https://www.icck.org/article/abs/TSCC.2026.664028
KW  - medical imaging
KW  - hybrid CNN-Transformer
KW  - multi-scale attention
KW  - semantic fusion
KW  - deep learning
AB  - Precise segmentation of colorectal polyps in colonoscopy images is essential for timely cancer diagnosis and prevention. Nevertheless, current segmentation methods contend with intrinsic variability in polyp appearance, differences in size, shape, and texture, while preserving computational efficiency necessary for clinical implementation. In this paper, we present a novel segmentation architecture that integrates scale-specific attention mechanisms within a hybrid CNN-Transformer backbone to address these challenges. Our model employs Coordinate Attention for high-resolution feature maps to preserve spatial details essential for boundary delineation, and Channel Attention for deep semantic features to enhance discriminative capacity. These representations are progressively integrated through a hierarchical decoder with specialized fusion modules: Semantic Fusion for high-level features, and Detail-Preserving Fusion for low-level features. The proposed architecture achieves state-of-the-art performance across five benchmark datasets, demonstrating superior generalization and robustness in challenging clinical scenarios.
SN  - 3068-9287
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Khan2026ScaleSpeci,
  author = {Ikram Majeed Khan and Wisal Khan},
  title = {Scale-Specific Visual Sensing for Colonoscopy Polyp Segmentation via Hybrid CNN-Transformer Attention},
  journal = {ICCK Transactions on Sensing, Communication, and Control},
  year = {2026},
  volume = {3},
  number = {2},
  pages = {109-123},
  doi = {10.62762/TSCC.2026.664028},
  url = {https://www.icck.org/article/abs/TSCC.2026.664028},
  abstract = {Precise segmentation of colorectal polyps in colonoscopy images is essential for timely cancer diagnosis and prevention. Nevertheless, current segmentation methods contend with intrinsic variability in polyp appearance, differences in size, shape, and texture, while preserving computational efficiency necessary for clinical implementation. In this paper, we present a novel segmentation architecture that integrates scale-specific attention mechanisms within a hybrid CNN-Transformer backbone to address these challenges. Our model employs Coordinate Attention for high-resolution feature maps to preserve spatial details essential for boundary delineation, and Channel Attention for deep semantic features to enhance discriminative capacity. These representations are progressively integrated through a hierarchical decoder with specialized fusion modules: Semantic Fusion for high-level features, and Detail-Preserving Fusion for low-level features. The proposed architecture achieves state-of-the-art performance across five benchmark datasets, demonstrating superior generalization and robustness in challenging clinical scenarios.},
  keywords = {medical imaging, hybrid CNN-Transformer, multi-scale attention, semantic fusion, deep learning},
  issn = {3068-9287},
  publisher = {Institute of Central Computation and Knowledge}
}