Department of Sociology, University of Hawai'i at Manoa.
Department of Electrical Engineering, University of Hawai'i at Manoa.
Melanoma Res. 2021 Dec 1;31(6):504-514. doi: 10.1097/CMR.0000000000000779.
Skin cancer remains the most commonly diagnosed cancer in the USA with more than 1 million new cases each year. Melanomas account for about 1% of all skin cancers and most skin cancer deaths. Multiethnic individuals whose skin is pigmented underestimate their risk for skin cancers and melanomas and may delay seeking a diagnosis. The use of artificial intelligence may help improve the diagnostic precision of dermatologists/physicians to identify malignant lesions. To validate our artificial intelligence's efficiency in distinguishing between images, we utilized 50 images obtained from our International Skin Imaging Collaboration dataset (n = 25) and pathologically confirmed lesions (n = 25). We compared the ability of our artificial intelligence to visually diagnose these 50 skin cancer lesions with a panel of three dermatologists. The artificial intelligence model better differentiated between melanoma vs. nonmelanoma with an area under the curve of 0.948. The three-panel member dermatologists correctly diagnosed a similar number of images (n = 35) as the artificial intelligence program (n = 34). Fleiss' kappa (ĸ) score for the raters and artificial intelligence indicated fair (0.247) agreement. However, the combined result of the dermatologists panel with the artificial intelligence assessments correctly identified 100% of the images from the test data set. Our artificial intelligence platform was able to utilize visual images to discriminate melanoma from nonmelanoma, using de-identified images. The combined results of the artificial intelligence with those of the dermatologists support the use of artificial intelligence as an efficient lesion assessment strategy to reduce time and expense in diagnoses to reduce delays in treatment.
皮肤癌仍是美国最常见的癌症，每年新发病例超过 100 万例。黑色素瘤约占所有皮肤癌和大多数皮肤癌死亡病例的 1%。皮肤色素沉着的多种族个体低估了自己患皮肤癌和黑色素瘤的风险，可能会延迟寻求诊断。人工智能的应用可能有助于提高皮肤科医生/医生诊断恶性病变的准确性。为了验证我们的人工智能在区分图像方面的效率，我们使用了来自我们的国际皮肤成像协作数据集（n=25）和病理证实病变（n=25）的 50 张图像。我们比较了我们的人工智能与三位皮肤科医生小组视觉诊断这 50 种皮肤癌病变的能力。人工智能模型在区分黑色素瘤与非黑色素瘤方面的曲线下面积为 0.948，更好地进行了区分。三位皮肤科医生小组成员正确诊断了与人工智能程序（n=34）相似数量的图像（n=35）。评分者和人工智能的 Fleiss' kappa（ĸ）评分表示为适度（0.247）一致。然而，皮肤科医生小组和人工智能评估的综合结果正确识别了测试数据集的所有图像（100%）。我们的人工智能平台能够使用经过身份验证的图像来区分黑色素瘤和非黑色素瘤，使用经过身份验证的图像。人工智能与皮肤科医生结果的综合结果支持将人工智能作为一种有效的病变评估策略，以减少诊断中的时间和费用，从而减少治疗延误。
