Skin Diseases Classification Using Deep Leaning Methods

Area of Science:

• Dermatology
• Computer Science
• Artificial Intelligence

Background:

• Skin tumors have a high incidence, necessitating advanced diagnostic tools.
• Current diagnosis relies on visual screening, dermoscopy, biopsy, and histopathology.
• Automatic classification of dermoscopic images is challenging due to subtle lesion variations.

Purpose of the Study:

• To propose a convolutional neural network (CNN) architecture for skin lesion classification.
• To utilize image pixels and diagnosis labels as inputs for the CNN model.
• To develop a powerful computer-aided diagnosis tool for dermatologists.

Main Methods:

• A CNN architecture was designed for skin lesion classification.
• The model was trained and validated on a public dataset of 10,015 skin lesion images.
• The dataset included seven distinct types of skin lesions.

Main Results:

• The CNN model demonstrated effectiveness in classifying various skin lesions.
• The approach leverages deep learning for feature extraction from dermoscopic images.
• The study validates the potential of CNNs in medical image classification tasks.

Conclusions:

• CNNs offer a promising approach for automated skin lesion classification.
• This method can serve as a valuable diagnostic aid for dermatologists.
• Further development can enhance the accuracy and scope of computer-aided dermatological diagnosis.