Skin Lesion Analysis and Cancer Detection Based on Machine/Deep Learning Techniques: A Comprehensive Survey

Area of Science:

• Dermatology
• Medical Imaging
• Artificial Intelligence

Background:

• Skin cancer, the most dangerous cancer, necessitates early diagnosis.
• Manual skin lesion examination is challenging and prone to errors.
• Computer-aided diagnosis (CAD) systems offer promising solutions for early skin cancer recognition.

Purpose of the Study:

• To conduct a comprehensive literature review of methodologies for skin lesion examination.
• To analyze existing techniques for skin cancer recognition.
• To identify challenges in current skin lesion analysis to guide future research.

Main Methods:

• Literature review encompassing preprocessing, segmentation, feature extraction, selection, and classification.
• Analysis of deep learning, machine learning, and computer vision approaches for skin lesion analysis.
• Examination of methodologies applied in skin cancer recognition.

Main Results:

• Computer-aided diagnosis approaches show impressive results in skin lesion analysis.
• Various techniques including deep learning and machine learning are utilized for skin cancer recognition.
• Despite advancements, challenges persist due to complex and rare skin lesion features.

Conclusions:

• Existing techniques for skin cancer discovery are effective but face challenges.
• Further research is needed to overcome obstacles in analyzing complex skin lesion features.
• Identifying these obstacles will aid researchers in advancing skin cancer detection.