Abstract
An AI-powered system for detecting skin lesions integrates IoMT platforms with CNN for feature extraction and Score-CAM for enhanced interpretability. The primary aim is to develop a system that accurately classifies skin lesions while providing visual justifications to boost clinician confidence, particularly in regions with limited dermatological resources. The system employs CNN-based feature extraction on dermoscopic images pre-processed using BM3D for noise reduction and CLAHE for contrast enhancement. Score-CAM integration improves model interpretability, while IoMT platforms ensure robustness and enable real-time diagnostics across diverse skin tones and lesion types. The proposed approach outperformed previous methods, achieving 99.20% accuracy, with real-time diagnostic capabilities offering significant benefits for remote and underserved areas. The model enhances accessibility, accuracy, and interpretability, making it highly beneficial for diverse populations.
References
- Mahmud, F., Mahfiz, M. M., Kabir, M. Z. I., & Abdullah, Y. (2023, December). An Interpretable Deep Learning Approach for Skin Cancer Categorization. In 2023 26th International Conference on Computer and Information Technology (ICCIT) Bangladesh, IEEE. 1-6
- Rahimiaghdam, Shakiba. "Enhancing the Stability and Quality Assessment of Visual Explanations for Thorax Disease Classification Using Deep Learning." (2023).
- Livieris, I. E., Pintelas, E., Kiriakidou, N., & Pintelas, P. (2023). Explainable Image Similarity: Integrating Siamese Networks and Grad-CAM. Journal of Imaging, 9(10), 224.
- Aasem, M., Iqbal, M. J., Ahmad, I., Alassafi, M. O., & Alhomoud, A. (2022). A survey on tools and techniques for localizing abnormalities in X-ray images using deep learning. Mathematics, 10(24), 4765.
- Manisarma, V., Sydulu, H. P., & Prashanthi, P. “AI-Driven Skin Disease Detection and Classification: Leveraging CNNs for Multiclass” Diagnosis. Journal of Cardiovascular Disease Research 13 (8) 3544-3553
- Bianchini, M., Andreini, P., & Bonechi, S. (2023). From Pixels to Diagnosis: AI-Driven Skin Lesion Recognition. In Advances in Smart Healthcare Paradigms and Applications: Outstanding Women in Healthcare—Volume 1 (pp. 115-135). Cham: Springer Nature Switzerland.
- Li, Z., Koban, K. C., Schenck, T. L., Giunta, R. E., Li, Q., & Sun, Y. (2022). Artificial intelligence in dermatology image analysis: current developments and future trends. Journal of clinical medicine, 11(22), 6826.
- Aggarwal, P., & Papay, F. A. (2022). Artificial intelligence image recognition of melanoma and basal cell carcinoma in racially diverse populations. Journal of Dermatological Treatment, 33(4), 2257-2262.
- Baskaran, D., Nagamani, Y., Merugula, S., & Premnath, S. P. (2023). MSRFNet for skin lesion segmentation and deep learning with hybrid optimization for skin cancer detection. The Imaging Science Journal, 71(7), 616-635.
- Nie, Y., Ferro, M., Sommella, P., Carratù, M., Cacciapuoti, S., Di Leo, G., ... & Fabbrocini, G. (2021, June). Ensembling CNNs for dermoscopic analysis of suspicious skin lesions. In 2021 IEEE International Symposium on Medical Measurements and Applications (MeMeA) IEEE. 1-6.
- Priyanka Pramila, R., & Subhashini, R. (2023). Automated skin lesion detection and lassification using fused deep convolutional neural network on dermoscopic images. Computational Intelligence, 39(6), 1073-1087.
- https://www.kaggle.com/datasets/mykeysid10/skin-lesions-dataset/