12 August 2022 in Science Advances
An estimated 3 billion people lack access to dermatological care globally. Artificial intelligence (AI) may aid in triaging skin diseases and identifying malignancies. However, most AI models have not been assessed on images of diverse skin tones or uncommon diseases. Thus, we created the Diverse Dermatology Images (DDI) dataset—the first publicly available, expertly curated, and pathologically confirmed image dataset with diverse skin tones. We show that state-of-the-art dermatology AI models exhibit substantial limitations on the DDI dataset, particularly on dark skin tones and uncommon diseases. We find that dermatologists, who often label AI datasets, also perform worse on images of dark skin tones and uncommon diseases. Fine-tuning AI models on the DDI images closes the performance gap between light and dark skin tones. These findings identify important weaknesses and biases in dermatology AI that should be addressed for reliable application to diverse patients and diseases.
1 January 2019 in Journal of the American Academy of Dermatology
CBChristopher K. Bichakjian
Christopher K. Bichakjian
CCClara Curiel-Lewandrowski
Clara Curiel-Lewandrowski
+10
1 October 2005 in Journal of Investigative Dermatology
16 December 2011 in Cancer
BACKGROUND.: Reduced melanoma mortality should result from an improved understanding of modifiable factors related to early detection. The authors of this report surveyed newly diagnosed patients to identify differences in prediagnosis behavioral and medical care factors associated with thinner versus thicker melanoma.METHODS.: In total, 566 adults with invasive melanoma completed questionnaires within 3 months of diagnosis on demographics, health care access, skin self‐examination (SSE), and physician skin examination (PSE) practices in the year before diagnosis. SSE was measured by us e of a melanoma picture aid and routine examination of some/all body sites versus none. Patient‐reported partial or full‐body PSE also was assessed. Melanoma thickness was dichotomized at 1 mm.RESULTS.: Patient ranged in age from 18 years to 99 years, and 61% were men. The median tumor thickness was 1.25 mm, and 321 tumors (57%) were >1 mm thick. Thinner tumors (≤1 mm) were associated with age ≤60 years (P = .0002), women (P = .0127), higher education level (P = .0122), and physician discovery (P ≤ .0001). Patients who used a melanoma picture aid and performed routine SSE were more likely to have thinner tumors than those who did not (odds ratio [OR], 2.66; 95% confidence interval [CI], 1.48‐4.80). Full‐body PSE was associated with thinner tumors (OR, 2.51; 95% CI, 1.62‐3.87), largely because of the effect of PSE in men aged >60 years (OR, 4.09 95% CI, 1.88‐8.89).CONCLUSIONS.: SSE and PSE were identified as complementary early detection strategies, particularly in men aged >60 years, in whom both partial and full‐body PSE were associated with thinner tumors. Given the high rates of physician access, PSE may be a more practical approach for successful early detection in this subgroup with highest mortality. Cancer 2012. © 2011 American Cancer Society.
7 January 2021 in Clinical Cancer Research
CCClara Curiel-Lewandrowski
Clara Curiel-Lewandrowski
+30
Five years ago, the Melanoma Research Foundation (MRF) conducted an assessment of the challenges and opportunities facing the melanoma research community and patients with melanoma. Since then, remarkable progress has been made on both the basic and clinical research fronts. However, the incidence, recurrence, and death rates for melanoma remain unacceptably high and significant challenges remain. Hence, the MRF Scientific Advisory Council and Breakthrough Consortium, a group that includes clinicians and scientists, reconvened to facilitate intensive discussions on thematic areas essential to melanoma researchers and patients alike, prevention, detection, diagnosis, metastatic dormancy and progression, response and resistance to targeted and immune-based therapy, and the clinical consequences of COVID-19 for patients with melanoma and providers. These extensive discussions helped to crystalize our understanding of the challenges and opportunities facing the broader melanoma community today. In this report, we discuss the progress made since the last MRF assessment, comment on what remains to be overcome, and offer recommendations for the best path forward.
1 August 2006 in Archives of Dermatology
Research from JAMA Dermatology — Facial Resurfacing for Nonmelanoma Skin Cancer Prophylaxis