Skin cancer is the most common form of cancer in the United States. Every fifth person has a chance of developing basal cell carcinoma (BCC), squamous cell carcinoma (SCC), melanoma, or other types of skin cancer before the age of […]
Skin cancer is the most common form of cancer in the United States. Every fifth person has a chance of developing basal cell carcinoma (BCC), squamous cell carcinoma (SCC), melanoma, or other types of skin cancer before the age of 70. Early detection of skin cancer is crucial to prevent deformities or loss of life. With the advent of artificial intelligence (AI) in various aspects of our lives, will AI become omnipresent in dermatology? Will it assist dermatologists in the early and faster detection of skin cancer? The answer is yes, eventually, it will become part of our future practice, but it is not quite ready yet.
Advantages of Applying Artificial Intelligence in the Diagnosis of Skin Conditions
Traditional methods of diagnosing skin cancer may become a thing of the past as artificial intelligence becomes more prevalent in supporting physicians. But how accurate are these methods? Can they harm patients, and will insurance plans accept and even require them? The goal of AI tools is to improve the diagnosis of skin cancer by using machine learning methods trained to detect and classify skin cancer using computer algorithms and deep neural networks. A recent analysis of AI, based on a systematic review, reveals the “strong potential of AI-based techniques in improving diagnostic accuracy and patient outcomes through early identification of melanoma.” It is also noted that further studies are needed to assess the applicability of these AI techniques to different populations and skin types, improve image processing standardization, and further compare the performance of AI techniques with board-certified dermatologists to evaluate clinical applicability.
Challenges in the Application of Artificial Intelligence in the Diagnosis of Skin Conditions
When it comes to artificial intelligence, we as physicians must ask ourselves, “How intelligent and accurate is it?” Many factors are necessary for the development of reliable sources of artificial intelligence, starting from the quality of data, size of the database, how they were annotated, who performed the data annotation, who trained the database, and whether the data or training were prone to biases. In the end, how much transparency is available, and how confident are we in AI? “Machine learning and deep learning are branches of artificial intelligence that deal with statistical modeling and inference, gradually learning from data inputs to predict desired goals and features.”
Artificial Intelligence and Available Tools in the Market
There are various AI tools available in the market:
1. MelaFind was the first tool approved by the FDA in 2011 but is no longer available. It was withdrawn in 2015 due to the lack of approval for the device user interface and the absence of probability and histogram data within the MelaFind device.
2. DermaSensor’s FDA submission for approval is currently under review and not available for sale in the United States. “DermaSensor is a portable objective skin cancer detection device that uses pulse and light and spectroscopy to non-invasively identify subcellular-level skin lesion information. This device utilizes Elastic Scattering Spectroscopy (ESS), which measures and records patterns of photon scattering as they bounce off different cellular structures after being exposed to rapid bursts of light.”
3. Nevisense is currently the only FDA-approved device for melanoma detection. It uses Electrical Impedance Spectroscopy, which is a “spectroscopic analysis of skin lesions that employs a software program to calculate and extract cell and skin structure information. The method uses a beam of light that penetrates beneath the skin’s surface. Light images captured by a digital camera or scanner are input into a computer for detailed analysis.”
4. Sklip is an AI skin cancer triage algorithm developed by dermatologists in the United States. Sklip dermoscopy algorithm received FDA breakthrough designation in 2021 and uses digital dermoscopy images and a dermoscopic three-point checklist to mark skin lesions as suspicious. This software as a medical device (SaMD) has completed national clinical studies for use in a professional environment with MD/DO/PA/NP healthcare providers.
5. MetaOptima is an AI skin cancer triage algorithm that received FDA breakthrough designation in 2021 and uses digital dermoscopy images for analysis.
6. SkinAnalytics currently has approval in the United Kingdom and uses machine learning to recognize benign, pre-cancerous, and common malignant lesions by analyzing images from the assessment system of over 70,000 NHS patients with suspected skin cancer.
Conclusion and Perspective on Artificial Intelligence in the Diagnosis of Skin Conditions
It is crucial to consider data quality and training, as well as the expertise of experts who developed AI tools. As the application of artificial intelligence in the diagnosis of skin conditions continues to evolve, standards need to be established for its implementation. Strong regulatory frameworks from governments and companies are needed to utilize the technology in the best interest of patients without causing harm. The good news is that the FDA strictly supervises skin cancer triage algorithms. If used properly, AI can become a powerful tool in practice. Dermatologists need to embrace the inevitability of artificial intelligence and consider it not as a threat but as an auxiliary aid in diagnosis rather than an independent entity.
Frequently Asked Questions (FAQ)
Q: How accurate are AI methods in diagnosing skin cancer?
A: According to recent analysis of AI methods based on machine learning, the results show “strong potential of AI-based techniques in improving diagnostic accuracy and patient outcomes through early detection of melanoma.”
Q: What advantages does the application of artificial intelligence bring to the diagnosis of skin cancer?
A: The application of artificial intelligence can help in the early detection of skin cancer, improvement of diagnostic accuracy and treatment outcomes. It can also enable faster recognition of suspicious lesions and better triage.
Q: What are the challenges in the application of artificial intelligence in the diagnosis of skin conditions?
A: Several factors can be obstacles in the application of artificial intelligence, including data quality and training, as well as regulatory and legal frameworks relating to the use of these technologies.
Q: What are the available AI tools for diagnosing skin cancer?
A: There are various AI tools available in the market such as MelaFind, DermaSensor, Nevisense, Sklip, MetaOptima, and SkinAnalytics.
Q: How can better implementation of artificial intelligence in the diagnosis of skin conditions be achieved?
A: To achieve better implementation of artificial intelligence, it is important to establish standards and regulations, ensure data quality and appropriate training, and involve expert input in the development and application of AI tools.