Deep Learning Framework For Skin Disease Detection Using Dynamic Filter, SAM Segmentation, YOLO Detection And Efficientnet-B5

Kavitha B; Dr. Kusuma Kumari B M

doi:10.29284/577qnz74

Authors

Kavitha B Research Scholar, Department of Studies and Research in Computer Applications,Tumkur University, Tumakuru, Karnataka
Dr. Kusuma Kumari B M Assistant Professor, Department of Studies and Research in Computer Applications,Tumkur University, Tumakuru, Karnataka

DOI:

https://doi.org/10.29284/577qnz74

Keywords:

Skin disease, HAM10000, dynamic filter, SAM, YOLO, classification.

Abstract

The reliable diagnosis and clinical decision making can highly depend on the early and accurate detection of the affected regions in medical images. The article displays a comprehensive Image Enhancement, Segmentation and Classification using publicly available datasets with the help of deep learning approaches. Our proposed framework takes large amounts of data i.e. around 10000 images. The scale of the dataset can increase this period. The first action in the workflow. Next, three different filters are applied and the best filter is selected using four quantitative image evaluation metrics and they are Peak Signal to noise Ratio (PSNR), Structural Similarity Index Measurement (SSIM), Mean Square Error (MSE), Edge Preserving Index (EPI). The improved visuals are then supplied to Segment Anything Model (SAM), which very precisely generates masks of affected region and retrieves the region in a very robust automatically. The masked results are further processed and compiled as YOLO (You Only Look Once) inputs to generated only affected region. For classification task, efficientnet-b5 deep learning architecture is used to deal with spatial complexities. The performance is evaluated based on accuracy (96.45%), precision (96.26), recall (96.17), and f1-score (96.19). Scripting language employed is python.

References

[1] Alam, T. M., et al. “An Efficient Deep Learning-Based Skin Cancer Classifier Using the HAM10000 Dataset.” Diagnostics, 2022.

[2] Behara, Kavita, Ernest Bhero, and John Terhile Agee. "An improved skin lesion classification using a hybrid approach with active contour snake model and lightweight attention-guided capsule networks." Diagnostics 14.6 (2024): 636.

[3] Debelee, Taye Girma. "Skin lesion classification and detection using machine learning techniques: a systematic review." Diagnostics 13.19 (2023): 3147.

[4] Gessert, N., et al. “Skin Lesion Classification Using Ensembles of Multi-Resolution EfficientNets.” Computers in Biology and Medicine, 2020.

[5] Jain, Satin, et al. "Deep learning-based transfer learning for classification of skin cancer." Sensors 21.23 (2021): 8142.

[6] Kassem, M. A., et al. "Machine learning and deep learning methods for skin lesion classification and diagnosis: a systematic review. Diagnostics. 2021; 11: 1390." 2021,

[7] Kreouzi, Magdalini, et al. "Deep learning for melanoma detection: a deep learning approach to differentiating malignant melanoma from benign melanocytic nevi." Cancers 17.1 (2024): 28.

[8] Naqvi, Maryam, et al. "Skin cancer detection using deep learning—a review." Diagnostics 13.11 (2023): 1911.

[9] Raghavendra, Paravatham VSP, et al. "Deep learning–based skin lesion multi-class classification with global average pooling improvement." Journal of Digital Imaging 36.5 (2023): 2227-2248.

[10] Shetty, Bhuvaneshwari, et al. "Skin lesion classification of dermoscopic images using machine learning and convolutional neural network." Scientific Reports 12.1 (2022): 18134.

[11] Tschandl, Philipp, Cliff Rosendahl, and Harald Kittler. "The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions." Scientific data 5.1 (2018): 180161.

[12] Wu, Yinhao et al. “Skin Cancer Classification With Deep Learning: A Systematic Review.” Frontiers in oncology vol. 12 893972. 13 Jul. 2022, doi:10.3389/fonc.2022.893972

[13] Alotaibi, Areej, and Duaa AlSaeed. “Skin Cancer Detection Using Transfer Learning and Deep Attention Mechanisms.” Diagnostics (Basel, Switzerland) vol. 15,1 99. 3 Jan. 2025, doi:10.3390/diagnostics15010099

[14] Akinrinade, Olusoji, and Chunglin Du. "Skin cancer detection using deep machine learning techniques." Intelligence-Based Medicine 11 (2025): 100191.

[15] Sara, Umme, Morium Akter, and Mohammad Shorif Uddin. "Image quality assessment through FSIM, SSIM, MSE and PSNR—a comparative study." Journal of Computer and Communications 7.3 (2019): 8-18.

[16] Mostafiz Ahammed, Md. Al Mamun, Mohammad Shorif Uddin, A machine learning approach for skin disease detection and classification using image segmentation, Healthcare Analytics, Volume 2, 2022, 100122, ISSN 2772-4425, https://doi.org/10.1016/j.health.2022.100122.

[17] Ajani, Atinuke Arinola, Ishaya Gambo, and Oluwasegun Faniyi. "Artificial Intelligence–Assisted Diagnosis of Atopic Eczema in Darker Skin Types: A CNN-Based Study." Dermatology Practical & Conceptual 16.1 (2026): 5897.

[18] Tanantong, Tanatorn, et al. "Mobile-Based Deep Learning Framework for Classifying Common Skin Diseases in Thailand." ICIC Express Letters Part B: Applications ICIC International 15.5 (2024): 495-503.

[19] Inthiyaz, Syed, et al. "Skin disease detection using deep learning." Advances in Engineering Software 175 (2023): 103361.