Leveraging Artificial Intelligence to Improve Diversity of Dermatological Skin Color Pathology: Protocol for an Algorithm Development and Validation Study.
Researchers
Journal
Modalities
Models
Abstract
The paucity of dark skin images in dermatological textbooks and atlases is a reflection of racial injustice in medicine. The underrepresentation of dark skin images makes diagnosing skin pathology in people of color challenging. In conditions when early diagnosis makes a difference between life and death, such as skin cancer, people of color have a worse prognosis and lower survival rates compared to others with lighter skin tones, as a result of delayed or incorrect diagnoses. Recent advances in artificial intelligence such as deep learning offer a potential solution by diversifying the mostly white skin image repositories and developing inclusive early diagnosis systems.We aim to develop and evaluate an artificial intelligence-based skin cancer early detection system for all skin tones using clinical images through generating diverse skin tone realistic images.This study consists of four phases: (1) publicly available skin image repositories will be analyzed to quantify the underrepresentation of the darker skin tones, (2) skin images will then be generated for the underrepresented tones, (3) the generated images will be extensively evaluated for realism and disease presentation through quantitative image quality assessment as well as qualitative human expert and non-expert ratings, and (4) the images will be utilized with available white skin images to develop a robust skin cancer early detection model.This study started in September 2020. The first phase of quantifying the underrepresentation of darker skin tones was completed in March 2021. The second phase of generating the images is in progress and will be completed by March 2022. The third phase is expected to be completed by May 2022, and the final phase is expected to be completed by September 2022.This work is the first step towards expanding skin tone diversity in existing image databases to address the current gap in the underrepresentation of darker skin tones. Once validated, the generated image bank will present a valuable resource that can potentially be utilized in physician education and other research applications. Furthermore, the generated images are expected to improve the generalizability of skin cancer detection. The developed model will thus assist family physicians and general practitioners in evaluating skin lesions’ severity and efficient triaging of the referrals to expert dermatologists for further assessment. In addition, the model can benefit dermatologists in diagnosing skin lesions.