Utilizing deep learning algorithms for the early identification and categorization of skin cancer
Junaid Iqbal1, Mohammad Faisal1, Subhan Ullah2
1Department of Computer Science, University of Malakand, Chakdara 18800, Pakistan.
Journal, Genetic Engineering & Biotechnology
|December 12, 2025
Summary
Early detection of malignant melanoma is crucial for effective skin cancer treatment. Deep learning models, particularly EfficientNet-B0, show high accuracy in identifying melanoma from images, paving the way for mobile health applications.
Area of Science:
- Dermatology and Medical Imaging
- Artificial Intelligence in Healthcare
Background:
- Skin cancer, particularly malignant melanoma, presents a significant global health challenge.
- Early detection of melanoma is critical for successful treatment outcomes.
- Image processing and deep learning offer promising avenues for early melanoma diagnosis.
Purpose of the Study:
- To evaluate the efficacy of deep learning models for early-stage melanoma detection.
- To compare the performance of EfficientNet-B0, VGG16, and Inception-V3 in classifying melanoma images.
- To explore the potential for a mobile platform integration for enhanced accessibility.
Main Methods:
- Utilized image preprocessing techniques.
- Trained EfficientNet-B0, VGG16, and Inception-V3 models on PH2 and ISIC datasets for 30 epochs.
- Assessed model performance based on accuracy metrics.
Main Results:
- All three models demonstrated strong performance in melanoma detection.
- EfficientNet-B0 achieved the highest accuracy at 92%.
- Inception-V3 and VGG-16 achieved accuracies of 87% and 85%, respectively.
Conclusions:
- Deep learning models are effective tools for early melanoma detection.
- EfficientNet-B0 shows superior performance among the tested models.
- Future work will focus on developing a mobile platform for on-the-go melanoma detection, improving accessibility and user adoption.
