Research on Dermoscopic Lesion Classification Based on Deformable Transition Layer and Multi-path Hybrid Attention Fusion

Yan Huang; Chaoan Cai

doi:10.54097/fpdnxa98

Authors

Yan Huang
Chaoan Cai

DOI:

https://doi.org/10.54097/fpdnxa98

Keywords:

Dermoscopic Image, Lesion Classification, DenseNet121, Deformable Convolution, Hybrid Attention Block

Abstract

As the largest organ of the human body, the skin undertakes crucial physiological functions such as protecting the organism and regulating metabolism, and its health status is directly related to the quality of human life. Skin lesions are diverse in types and similar in morphology; failure to accurately identify and intervene in a timely manner can easily lead to delayed diagnosis and treatment, posing a serious threat to patients' lives and health. However, dermoscopic images are commonly plagued by irregular lesion morphology, blurred boundaries and low discrimination of fine-grained features, which restrict the classification performance of traditional deep learning models. To address the above pain points, an improved DenseNet121 architecture integrating multi-path hybrid attention and deformable transition layer (HADT-DenseNet) is proposed. Based on DenseNet121, this architecture introduces deformable convolution to replace the traditional convolution in the transition layer to realize adaptive sampling of irregular lesion features. Meanwhile, a multi-path branch structure is designed, with hybrid attention blocks inserted between key dense blocks to enhance attention features in both channel and spatial dimensions. Through a hierarchical feature fusion strategy, the basic dense features are organically combined with attention-enhanced features to improve the multi-level nature of feature expression. Experimental results on the public ISIC 2017 dataset show that HADT-DenseNet achieves a classification accuracy of 92.3%, a precision of 88.2%, an AUC of 96.1% and an F1-score of 91.8%, which is significantly superior to the baseline model DenseNet121 and other improved variants based on common attention mechanisms. This study provides an effective and feasible new approach for the intelligent auxiliary diagnosis of skin lesions.

Downloads

Download data is not yet available.

References

[1] Arnold M, Singh D, Laversanne M, et al. Global burden of cutaneous melanoma in 2020 and projections to 2040[J]. JAMA Dermatology. 2022, 158(5): 495-503.

[2] Flohr C, Hay R. Putting the burden of skin diseases on the global map[J]. British Journal of Dermatology. 2021, 184(2): 189-190.

[3] Naqvi S A R, Mobashsher A T, Mohammed B, et al. Benign and malignant skin lesions: Dielectric characterization, modelling and analysis in frequency band 1 to 14 GHz[J]. IEEE Transactions on Biomedical Engineering. 2022, 70(2): 628-639.

[4] Kumari S, Choudhary P K, Shukla R, et al. Recent advances in nanotechnology based combination drug therapy for skin cancer[J]. Journal of Biomaterials Science, Polymer Edition. 2022, 33(11): 1435-1468.

[5] Alshmrani S A, Alotaibi M F, Alfakeeh S A. Fusion of multi CNN features with ANN for early classification of melanoma using dermoscopy images[J]. Discover Sustainability. 2026, 7(1): 211-211.

[6] Alhassan M A, Altmami I N. DL-PCMNet: Distributed Learning Enabled Parallel Convolutional Memory Network for Skin Cancer Classification with Dermatoscopic Images[J]. Diagnostics. 2026, 16(2): 359-359.

[7] Cho J, Shanmugavadivel K, Subramanian M, et al. Attention-aware Deep Learning Models for Dermoscopic Image Classification for Skin Disease Diagnosis[J]. Current Medical Imaging. 2025, 21

[8] Naeem A, Malik H, Din Z M, et al. SkinDWNet: a novel deep learning model for multiclass classification of skin cancers using dermoscopic images[J]. Multimedia Systems. 2025, 31(4): 314-314.

[9] Din Z M, Naeem A, Malik H, et al. Skin cancer classification using a borderline-SMOTE enhanced neural network model on dermoscopic images[J]. Biomedical Signal Processing and Control. 2026, 118: 109691-109691.

[10] Dai J, Qi H, Xiong Y, et al. Deformable convolutional networks[C]. Proceedings of the IEEE International Conference on Computer Vision. 2017: 764-773.

[11] Huang G, Liu Z, Laurens V D M, et al. Densely connected convolutional networks[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017: 4700-4708.

[12] Hu J, Shen L, Sun G. Squeeze-and-Excitation Networks[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE. 2018: 7132-7141.

[13] Woo S, Park J, Lee J Y, et al. CBAM: Convolutional Block Attention Module[C]. Computer Vision-ECCV 2018. Cham: Springer, 2018: 3-19.

[14] Chen X , Wang X , Zhou J , et al. Activating More Pixels in Image Super-Resolution Transformer[C]. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2023: 22367-22377.

[15] Codella N C F, Gutman D, Celebi M E, et al. Skin lesion analysis toward melanoma detection: A challenge at the 2017 International Symposium on Biomedical Imaging (ISBI), hosted by the International Skin Imaging Collaboration (ISIC)[J]. IEEE Transactions on Medical Imaging, 2018, 37(11): 2642-2659.