Explainable Multi-Backbone Deep Learning with Attention-Based Feature Fusion Framework for Laryngeal Carcinoma Classification using Medical Imaging
International Journal of Medical Toxicology and Forensic Medicine,
Vol. 16 (2026),
1 January 2026
,
Page 1-7
https://doi.org/10.22037/ijmtfm.v16.53090
Abstract
Background: Automated diagnostic classification of laryngeal carcinoma using laryngoscopic endoscopic images can ease the workload of otolaryngologists, thereby enhancing their efficiency in practice and supporting young clinicians in achieving accurate diagnoses. Deep Learning has become an effective approach for medical image classification and relies on large labeled datasets. However, several existing methods for laryngeal carcinoma analysis often struggle to capture local and global dependencies, thereby affecting classification accuracy and reliability.
Methods: This study presents an Explainable Multi-Backbone Deep Learning Framework for Laryngeal Carcinoma Detection (XMBDL-LCD) in Medical Imaging. The proposed technique utilizes ConvNeXt V2, Swin Transformer V2, and DINOv2 models to extract complementary feature representations from laryngeal images. Besides, the proposed method uses a Multi-Head Cross-Attention Fusion mechanism to combine the extracted features into a unified representation. A Kolmogorov-Arnold Networks-based classifier is applied to perform the final classification of laryngeal carcinoma. To enhance the performance of the proposed method, Explainable AI using Grad-CAM++ is incorporated to visualize the regions that contribute to the model's decision.
Results: The proposed XMBDL-LCD method can be evaluated on a laryngeal endoscopic dataset, and the results demonstrate its efficiency compared with other recent medical image classification approaches, with a maximum accuracy of 94.57%.
Conclusion: Therefore, the proposed model is an assistive tool for laryngeal carcinoma using laryngoscopic endoscopic images.
- Laryngeal Carcinoma Classification, Medical Image Analysis, Deep Learning, Attention-Based Feature Fusion, Kolmogorov-Arnold Networks, GradCAM++
How to Cite
References
[1] Wang SS, Cheng WJ, Wang CT, Hung CH, Fang SH. Improved l. Expert Syst Appl a critical history-based multimodal approach. Expert Syst Appl. 2026:132643. [DOI: 10.1016/j.eswa.2026.132643]
[2] Kim HB, Song J, Park S, Lee YO. Classification of laryngeal diseases including laryngeal cancer, benign mucosal disease, and vocal cord paralysis by artificial intelligence using voice analysis. Sci Rep. 2024;14(1):9297. [DOI: 10.1038/s41598-024-58817-x]
[3] Kang YF, Yang L, Hu YF, Xu K, Cai LJ, Hu BB, et al. Self-attention mechanisms-based laryngoscopy image classification technique for laryngeal cancer detection. Head Neck. 2025;47(3):944-55. [DOI: 10.1002/hed.27999]
[4] Chen X, Xia S, Jiang H, Lv F, Yu Q, Ning Y, et al. CT-based radiomics and deep learning models for predicting thyroid cartilage invasion and patient prognosis in laryngeal carcinoma. Sci Rep. 2025;15(1):40193. [DOI: 10.1038/s41598-025-23809-y]
[5] Al Khulayf AMF, Alamgeer M, Obayya M, Alqahtani M, Ebad SA, Aljehane NO, et al. Mathematical fusion of efficient transfer learning with fine-tuning model for laryngeal cancer detection and classification in the throat region using biomedical images. Alexandria Eng J. 2025;129:658-71. [DOI: 10.1016/j.aej.2025.06.044]
[6] Jiang H, Xie K, Chen X, Ning Y, Yu Q, Lv F, et al. A prognostic model integrating radiomics and deep learning based on CT for survival prediction in laryngeal squamous cell carcinoma. Sci Rep. 2025;15(1):29990. [DOI: 10.1038/s41598-025-15166-7]
[7] Gu M, Yu B, Liu H, Kong H. Multivariate feature analysis of early-stage laryngeal cancer serum components using surface-enhanced Raman spectroscopy. Spectrochim Acta A Mol Biomol Spectrosc. 2026:127625. [DOI: 10.1016/j.saa.2026.127625]
[8] Nie X, Zhang X, Wang D, Liu Y, Xing L, Liu W. Laryngeal cancer diagnosis based on improved YOLOv8 algorithm. Mach Learn Sci Technol. 2025;6(1):015011. [DOI: 10.1088/2632-2153/ada2d9]
[9] Srivastava R, Kumar N, Sandhan T. Binary classification of laryngeal images utilising ResNet-50 CNN architecture. Indian J Otolaryngol Head Neck Surg. 2025;77(2):644-51. [DOI: 10.1007/s12070-024-05202-9]
[10] Alazwari S, Maashi M, Alsamri J, Alamgeer M, Ebad SA, Alotaibi SS, et al. Improving laryngeal cancer detection using chaotic metaheuristics integration with squeeze-and-excitation ResNet model. Health Inf Sci Syst. 2024;12(1):38. [DOI: 10.1007/s13755-024-00296-5]
[11] Sachane MN, Patil SA. Hybrid QDCNN-DNFN for laryngeal cancer detection using image and voice analysis in federated learning. Sens Imaging. 2024;26(1):1. [DOI: 10.1007/s11220-024-00531-z]
[12] Huo G, Zhang Y, Gao J, Wang B, Hu Y, Yin B. CaEGCN: cross-attention fusion based enhanced graph convolutional network for clustering. IEEE Trans Knowl Data Eng. 2023;35(4):3471-83. [DOI: 10.1109/TKDE.2021.3125020]
[13] Sait ARW, AlBalawi E, Nagaraj R. Ensemble learning driven Kolmogorov-Arnold networks-based lung cancer classification. PLoS One. 2024;19(12):e0313386. [DOI: 10.1371/journal.pone.0313386]
[14] BR B. Comparative analysis of Grad-CAM variants for enhancing explainability in PCOS detection using ultrasound imaging. 2025. [DOI: 10.12785/ijcds/1571152580]
[15] Laryngeal dataset [dataset on the Internet]. Kaggle; [date unknown] [cited 2026 Sep 21]. [Link]
[16] Emre E, Cetintas D, Yildirim M, Emre S. Vocal fold disorders classification and optimization of a custom video laryngoscopy dataset through structural similarity index and a deep learning-based approach. J Clin Med. 202 structural similarity index and a deep learning-based approach. J Clin Med. [DOI: 2023390/jcm14196899]
[17] Xu ZH, Fan DG, Huang JQ, Wang JW, Wang Y, Li YZ. Computer-aided diagnosis of laryngeal cancer based on deep learning with laryngoscopic images. Diagnostics. 2023;13(24):3669. [DOI: 10.3390/diagnostics13243669]
- Abstract Viewed: 0 times
- 53090 Downloaded: 0 times