A comprehensive review of artificial intelligence methods and applications in skin cancer diagnosis and treatment: Emerging trends and challenges

Abstract

A substantial body of research has been published on artificial intelligence applications in skin cancer influenced by the latter's rising rates, the scarcity of specialized healthcare professionals, and rapid advancements in automated diagnosis and treatment methods. We present a comprehensive review employing text mining to identify key themes of artificial intelligence in skin cancer diagnosis and treatment research. Our text mining model uncovers nine key topics, including dermatological data, machine and deep learning methods, segmentation, data generation, melanoma, basal cell carcinoma, model validation, and treatment. We extensively review the literature on each topic to offer valuable insights and highlight research gaps. Our findings indicate a need for a comprehensive and diverse dataset that includes lesion images, clinical data, and treatment information. In addition, our topic analysis ranks deep learning-based diagnosis as the top topic, followed by data generation and melanoma diagnosis. These insights demonstrate the bias towards deep learning methods and the shortage of studies on rare and precancerous skin lesions. Despite the gaps defined, artificial intelligence can be utilized for triage, initial screening, providing second opinion in diagnosing complex cases, and educational purposes. Additionally, artificial intelligence models can enhance patient outcomes through early diagnosis, treatment recommendation, and response prediction.