文献速递骨软组织肉瘤与黑色素瘤

卷积神经网络在多分类皮肤癌图像分类中系统性优于112名皮肤科医生

Systematic outperformance of 112 dermatologists in multiclass skin cancer image classification by convolutional neural networks.
PubMed 文献 · Eur J Cancer · 2019-08-14

本研究使用11,444张皮肤镜图像训练卷积神经网络,并在300张活检验证图像上比较其与112名德国皮肤科医生的分类性能。主要终点为良恶性分类正确率,次要终点为五种诊断类别的正确分类。结果显示,在同等灵敏度下算法特异性显著更高(91.3% vs 59.8%),多分类任务中除基底细胞癌外均显著优于医生。

要点

  • 在同等灵敏度下,CNN对良恶性分类的特异性达91.3%,显著高于医生的59.8%(p<0.001)。
  • 多分类任务中,CNN在同等灵敏度下特异性为98.8%,除基底细胞癌外均显著优于医生。
  • 研究表明皮肤镜图像自动分类可扩展至多分类问题,更贴近临床鉴别诊断需求。

核心结论

  • 卷积神经网络在多分类皮肤癌图像分类中系统性优于皮肤科医生,且性能显著(p<0.001)。
  • 自动化分类可从二分类扩展至多分类,更好地反映临床鉴别诊断。

研究设计与样本

  • 使用11,444张皮肤镜图像训练卷积神经网络,覆盖皮肤癌筛查中常见的色素性皮损诊断类别。
  • 测试集为300张活检验证图像,与来自13家德国大学医院的112名皮肤科医生进行比较。

主要发现

  • 医生对主要终点(良恶性分类)的灵敏度为74.4%(95%CI 67.0-81.8%),特异性为59.8%(95%CI 49.8-69.8%)。
  • 在同等灵敏度下,算法对主要终点的特异性为91.3%(95%CI 85.5-97.1%)。
  • 医生对次要终点(五分类)的平均灵敏度为56.5%(95%CI 42.8-70.2%),特异性为89.2%(95%CI 85.0-93.3%)。
  • 在同等灵敏度下,算法对次要终点的特异性为98.8%。
  • 主要终点双侧McNemar检验p<0.001;次要终点除基底细胞癌(表现相当)外,均p<0.001。

临床怎么用

  • CNN可辅助皮肤癌筛查中的多分类鉴别诊断,在保持高灵敏度同时显著提高特异性。
  • 对基底细胞癌的分类表现与医生相当,其余类别优于医生,提示可优先用于非基底细胞癌的鉴别。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND: Recently, convolutional neural networks (CNNs) systematically outperformed dermatologists in distinguishing dermoscopic melanoma and nevi images. However, such a binary classification does not reflect the clinical reality of skin cancer screenings in which multiple diagnoses need to be taken into account. METHODS: Using 11,444 dermoscopic images, which covered dermatologic diagnoses comprising the majority of commonly pigmented skin lesions commonly faced in skin cancer screenings, a CNN was trained through novel deep learning techniques. A test set of 300 biopsy-verified images was used to compare the classifier's performance with that of 112 dermatologists from 13 German university hospitals. The primary end-point was the correct classification of the different lesions into benign and malignant. The secondary end-point was the correct classification of the images into one of the five diagnostic categories. FINDINGS: Sensitivity and specificity of dermatologists for the primary end-point were 74.4% (95% confidence interval [CI]: 67.0-81.8%) and 59.8% (95% CI: 49.8-69.8%), respectively. At equal sensitivity, the algorithm achieved a specificity of 91.3% (95% CI: 85.5-97.1%). For the secondary end-point, the mean sensitivity and specificity of the dermatologists were at 56.5% (95% CI: 42.8-70.2%) and 89.2% (95% CI: 85.0-93.3%), respectively. At equal sensitivity, the algorithm achieved a specificity of 98.8%. Two-sided McNemar tests revealed significance for the primary end-point (p < 0.001). For the secondary end-point, outperformance (p < 0.001) was achieved except for basal cell carcinoma (on-par performance). INTERPRETATION: Our findings show that automated classification of dermoscopic melanoma and nevi images is extendable to a multiclass classification problem, thus better reflecting clinical differential diagnoses, while still outperforming dermatologists at a significant level (p < 0.001).

来源:Eur J Cancer,查看原文。版权归原作者所有。

引用

  1. Maron RC, Weichenthal M, Utikal JS, et al. Systematic outperformance of 112 dermatologists in multiclass skin cancer image classification by convolutional neural networks. Eur J Cancer. 2019. PMID: 31419752. doi:10.1016/j.ejca.2019.06.013

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。