Boosted dipper throated optimization algorithm-based Xception neural network for skin cancer diagnosis: An optimal...

Area of Science:

• Dermatology
• Artificial Intelligence
• Medical Imaging

Background:

• Skin cancer diagnosis relies on methods that are often invasive, time-consuming, or lack reliability.
• There is a critical need for advanced, automated, and non-invasive diagnostic tools for early skin cancer detection.

Purpose of the Study:

• To develop and validate a novel, automated, non-invasive method for accurate skin cancer diagnosis.
• To enhance the performance of deep learning models for skin lesion classification using optimization algorithms.

Main Methods:

• Utilized an Xception neural network, a deep learning model, for high-level feature extraction from dermoscopy images.
• Optimized the Xception network parameters using the Boosted Dipper Throated Optimization (BDTO) algorithm, a bio-inspired technique.
• Employed image preprocessing and data augmentation on the ISIC dataset to improve image quality and model generalization.

Main Results:

• The proposed method achieved high diagnostic performance, with an average precision of 94.936%, accuracy of 94.206%, and recall of 97.092%.
• Demonstrated superior performance compared to several contemporary skin cancer detection approaches.
• Validated the method's robustness and superiority through 5-fold ROC curve and error curve analysis.

Conclusions:

• The BDTO-optimized Xception neural network offers a highly accurate and efficient non-invasive approach for skin cancer diagnosis.
• This AI-driven method presents a significant advancement over existing diagnostic techniques, improving patient outcomes.
• The study highlights the potential of optimized deep learning models in revolutionizing dermatological diagnostics.