Efficient Attention-Guided MobileNet V2 with Explainable AI for Multi-Class Skin Disease Classification on HAM10000
DOI:
https://doi.org/10.30871/jaic.v10i3.13010Keywords:
class imbalance; convolutional block attention module; focal loss; MobileNetV2; skin lesion classificationAbstract
The increasing global incidence of skin cancer, particularly melanoma, coupled with a scarcity of dermatologists, necessitates the development of accurate and accessible AI-driven diagnostic tools. However, deep learning models often struggle with severe class imbalance in public dermoscopic datasets, leading to poor performance on minority lesion types. This research aims to enhance the diagnostic precision of a lightweight MobileNetV2 architecture for multi-class skin lesion classification by integrating the Convolutional Block Attention Module (CBAM) and employing Focal Loss. The methodology involves evaluating four model variants (Baseline, Baseline+CBAM, Baseline+Focal Loss, and Baseline+CBAM+Focal Loss) on the HAM10000 dataset, with performance measured by accuracy, precision, recall, and F1-score. The optimal model (M4) successfully achieved convergence without overfitting, demonstrating exceptional F1-scores for six of seven classes, including near-perfect classification for melanoma (0.96) and dermatofibroma (0.97). The primary limitation was the actinic keratosis class (F1-score 0.60) due to high morphological similarity with other lesions. In conclusion, the synergistic combination of CBAM and Focal Loss effectively mitigates class imbalance and enhances feature representation in a computationally efficient model, providing a robust and interpretable solution for skin cancer screening.
Downloads
References
[1] A. Alhudhaif, B. Almaslukh, A. O. Aseeri, O. Guler, dan K. Polat, “A novel nonlinear automated multi-class skin lesion detection system using soft-attention based convolutional neural networks,” Chaos, Solitons and Fractals, vol. 170, 2023, doi: 10.1016/j.chaos.2023.113409.
[2] G. Priyanka, D. Dhanabal, D. Divya, M. Hemanth, dan V. Karthika, “Skin Disease Detection Using Convolutional Neural Network,” 10th Int. Conf. Adv. Comput. Commun. Syst. ICACCS 2024, hal. 1949–1954, 2024, doi: 10.1109/ICACCS60874.2024.10716893.
[3] P. N. Srinivasu, J. G. SivaSai, M. F. Ijaz, A. K. Bhoi, W. Kim, dan J. J. Kang, “Classification of Skin Disease Using Deep Learning Neural Networks with MobileNet V2 and LSTM,” Sensors, vol. 21, no. 8, hal. 2852, Apr 2021, doi: 10.3390/s21082852.
[4] D. Penyakit, K. Dengan, dan M. Convolutional, “Jurnal Teknologi Terpadu Network Menggunakan Arsitektur VGG19,” J. Teknol. Terpadu, vol. 11, no. 2, hal. 87–93, 2025.
[5] M. M. Siregar, R. Hizria, dan D. Pardede, “Perbandingan Kinerja Kernel SVM dalam Klasifikasi Kategori Kanker Kulit Menggunakan Transfer Learning,” Data Sci. Indones., vol. 4, no. 1, hal. 83–90, 2024, doi: 10.47709/dsi.v4i1.4665.
[6] P. A. Prayesy, “Studi Perbandingan Metode Support Vector Machine, Random Forest, Dan Convolutional Neural Network Untuk Klasifikasi Penyakit Kulit,” J. Kecerdasan Buatan dan Teknol. Inf., vol. 4, no. 1, hal. 70–76, 2025, doi: 10.69916/jkbti.v4i1.214.
[7] G. Putra, H. Puja, E. Haerani, dan F. Syafria, “Implementation of Convolutional Neural Network Algorithm ( ResNet-50 ) for Benign and Malignant Skin Cancer Classification Implementasi Algoritma Convolutional Neural Network ( Resnet-50 ) untuk Klasifikasi Kanker Kulit Benign dan Malignant,” MALCOM Indones. J. Mach. Learn. Comput. Sci., vol. 4, no. July, hal. 984–992, 2024.
[8] I. Iqbal, M. Younus, K. Walayat, M. U. Kakar, dan J. Ma, “Automated multi-class classification of skin lesions through deep convolutional neural network with dermoscopic images,” Computerized Medical Imaging and Graphics, vol. 88. Elsevier BV, hal. 101843, 2021. doi: 10.1016/j.compmedimag.2020.101843.
[9] B. Shetty, R. Fernandes, A. P. Rodrigues, R. Chengoden, S. Bhattacharya, dan K. Lakshmanna, “Skin lesion classification of dermoscopic images using machine learning and convolutional neural network,” Scientific Reports, vol. 12, no. 1. Springer Science and Business Media LLC, 2022. doi: 10.1038/s41598-022-22644-9.
[10] H. Wu, J. Pan, Z. Li, Z. Wen, dan J. Qin, “Automated Skin Lesion Segmentation Via an Adaptive Dual Attention Module,” IEEE Transactions on Medical Imaging, vol. 40, no. 1. Institute of Electrical and Electronics Engineers (IEEE), hal. 357–370, 2021. doi: 10.1109/tmi.2020.3027341.
[11] S. Roy, R. R. Cherish, dan G. Roy, “An attention-based loss function and synthetic minority oversampling technique for alleviating class imbalance in predicting diabetes,” Healthcare Analytics, vol. 7. Elsevier BV, hal. 100399, 2025. doi: 10.1016/j.health.2025.100399.
[12] A. V, V. Commuri, V. Krishnan, dan P. T. N, “A Comprehensive framework for classifying skin lesion diseases with class imbalance handling,” 2025 International Conference on Biomedical Engineering and Sustainable Healthcare (ICBMESH). IEEE, hal. 1–5, 2025. doi: 10.1109/icbmesh66209.2025.11182197.
[13] G. S. Navya dan K. P. Rao, “Hybrid EfficientNetB3 and DenseNet201 with CBAM Attention for Multi-Class Skin Disease Classification,” 2025 3rd DMIHER International Conference on Artificial Intelligence in Healthcare, Education and Industry (IDICAIHEI). IEEE, hal. 1–5, 2025. doi: 10.1109/idicaihei65991.2025.11379054.
[14] R. O. Ogundokun et al., “Enhancing Skin Cancer Detection and Classification in Dermoscopic Images through Concatenated MobileNetV2 and Xception Models,” Bioengineering, vol. 10, no. 8. MDPI AG, hal. 979, 2023. doi: 10.3390/bioengineering10080979.
[15] V. Ravi, “Attention Cost-Sensitive Deep Learning-Based Approach for Skin Cancer Detection and Classification,” Cancers, vol. 14, no. 23. MDPI AG, hal. 5872, 2022. doi: 10.3390/cancers14235872.
[16] O. Salih dan K. J. Duffy, “Optimization Convolutional Neural Network for Automatic Skin Lesion Diagnosis Using a Genetic Algorithm,” Applied Sciences, vol. 13, no. 5. MDPI AG, hal. 3248, 2023. doi: 10.3390/app13053248.
[17] K. Behara, E. Bhero, dan J. T. Agee, “Skin Lesion Synthesis and Classification Using an Improved DCGAN Classifier,” Diagnostics, vol. 13, no. 16. MDPI AG, hal. 2635, 2023. doi: 10.3390/diagnostics13162635.
[18] D. Popescu, M. El-khatib, dan L. Ichim, “Skin Lesion Classification Using Collective Intelligence of Multiple Neural Networks,” Sensors, vol. 22, no. 12. MDPI AG, hal. 4399, 2022. doi: 10.3390/s22124399.
[19] T. H. H. Aldhyani, A. Verma, M. H. Al-Adhaileh, dan D. Koundal, “Multi-Class Skin Lesion Classification Using a Lightweight Dynamic Kernel Deep-Learning-Based Convolutional Neural Network,” Diagnostics, vol. 12, no. 9. MDPI AG, hal. 2048, 2022. doi: 10.3390/diagnostics12092048.
[20] V. D. Nguyen, N. D. Bui, dan H. K. Do, “Skin Lesion Classification on Imbalanced Data Using Deep Learning with Soft Attention,” Sensors, vol. 22, no. 19. MDPI AG, hal. 7530, 2022. doi: 10.3390/s22197530.
[21] S. M. Thwin dan H.-S. Park, “Skin Lesion Classification Using a Deep Ensemble Model,” Applied Sciences, vol. 14, no. 13. MDPI AG, hal. 5599, 2024. doi: 10.3390/app14135599.
[22] S. Marison, S. Silvanus, dan R. Rusdiah, “Ai-Based Algorithms for Network Security: Trends, Per-Formance, and Challenges,” JURTEKSI (Jurnal Teknol. dan Sist. Informasi), vol. 11, no. 2, hal. 329–336, 2025, doi: 10.33330/jurteksi.v11i2.3699.
[23] S. R. Shegar dan S. S. Patil, “Multi-Class Skin Lesion Classification Using Transfer Learning with EfficientNet-B3 and Convolutional Block Attention Module,” Journal of Smart Sensors and Computing, vol. 1, no. 3. GR Scholastic LLP, 2025. doi: 10.64189/ssc.25213.
[24] M. Kahia, B. Bassem, I. Sekkiou, dan F. Kallel, “Modified Multi-Head Attention Transformer (MMHAT) for Skin Image Classification.” MDPI AG, 2025. doi: 10.20944/preprints202501.1723.v1.
[25] Y. Zhang, X. Zhang, dan W. Zhu, “ANC: Attention Network for COVID-19 Explainable Diagnosis Based on Convolutional Block Attention Module,” Computer Modeling in Engineering & Sciences, vol. 127, no. 3. Tech Science Press, hal. 1037–1058, 2021. doi: 10.32604/cmes.2021.015807.
[26] Y. Fang, H. Huang, W. Yang, X. Xu, W. Jiang, dan X. Lai, “Nonlocal convolutional block attention module VNet for gliomas automatic segmentation,” International Journal of Imaging Systems and Technology, vol. 32, no. 2. Wiley, hal. 528–543, 2021. doi: 10.1002/ima.22639.
[27] [27] M. Shafiq, K. Aggarwal, J. Jayachandran, G. Srinivasan, R. Boddu, dan A. Alemayehu, “RETRACTED: A novel Skin lesion prediction and classification technique: ViT‐GradCAM,” Skin Research and Technology, vol. 30, no. 9. Wiley, 2024. doi: 10.1111/srt.70040.
[28] S. Deng et al., “A Real-time Lithological Identification Method based on SMOTE-Tomek and ICSA Optimization,” Acta Geol. Sin. (English Ed., vol. 98, no. 2, hal. 518–530, 2024, doi: 10.1111/1755-6724.15144.
[29] J. Padhye, V. Firoiu, &D. Towsley, “A stochastic model of TCP Reno congestion avoidance and control,” Univ. of Massachusetts, Amherst, MA, CMPSCI Tech. Rep. 99-02, 199
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Devi Larasati, Ucta Pradema Sanjaya
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License (Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) ) that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
