As artificial intelligence (AI) rapidly advances, its integration into dermatology, particularly through Generative Adversarial Networks (GANs), is opening new horizons in patient education and skin cancer diagnosis.
A possible application of GANs is the transformation of benign dermoscopic images into their artificially created malignant counterparts. This technology allows dermatologists to visually demonstrate to patients the subtle differences between benign and malignant lesions using actual dermoscopic images from the patient's body.
Figure 1 On the left, the original benign skin lesion; and on the right, the AI-generated image depicting the lesion's potential malignant progression.
During skin screening procedures, dermatologists can present both the original benign image and the AI-generated malignant version. This side-by-side comparison aids in explaining why certain lesions do not require excision and what changes to look for in the future. Conversely, for lesions appearing malignant, the AI can generate a benign counterpart to illustrate the necessity of excision.
Utilizing frame interpolation, we create a seamless and gradual transition from the original benign dermoscopic image to the AI-generated malignant counterpart. This results in a brief video, illustrating the subtle progression of skin changes, potentially enhancing patient understanding of skin cancer evolution.
This visual tool goes beyond the traditional ABCDE rule for skin cancer diagnosis, offering a more intuitive and understandable approach. By seeing the potential progression from benign to malignant, patients gain a clearer understanding of their diagnosis, fostering confidence in self-examinations and early detection.
Generative AI in dermatology is not just a technological advancement; it could be a step towards empowering patients with a deeper understanding of their skin health. By bridging the gap between complex medical information and patient comprehension, technology could enhance proactive skin care and early cancer detection.
Lennart Jütte (Leibniz University Hannover)