Skin cancer is a serious condition that requires accurate identification and
treatment. One way to assist clinicians in this task is by using computer-aided
diagnosis (CAD) tools that can automatically segment skin lesions from
dermoscopic images. To this end, a new adversarial learning-based framework
called EGAN has been developed. This framework uses an unsupervised generative
network to generate accurate lesion masks. It consists of a generator module
with a top-down squeeze excitation-based compound scaled path and an asymmetric
lateral connection-based bottom-up path, and a discriminator module that
distinguishes between original and synthetic masks. Additionally, a
morphology-based smoothing loss is implemented to encourage the network to
create smooth semantic boundaries of lesions. The framework is evaluated on the
International Skin Imaging Collaboration (ISIC) Lesion Dataset 2018 and
outperforms the current state-of-the-art skin lesion segmentation approaches
with a Dice coefficient, Jaccard similarity, and Accuracy of 90.1%, 83.6%, and
94.5%, respectively. This represents a 2% increase in Dice Coefficient, 1%
increase in Jaccard Index, and 1% increase in Accuracy.

使用对抗学习的算法开发的可自动生成皮肤病变掩膜的 EGAN 算法，可应用于皮肤癌病变的计算机辅助诊断，其精度比当前最先进的基于皮损分割的方法高出 2% 的 Dice 系数，1% 的 Jaccard 相似度，和 1% 的准确度。

基于生成对抗网络的皮损分割

Generative Adversarial Networks based Skin Lesion Segmentation

Convolutional Neural Networks have demonstrated human-level performance in
the classification of melanoma and other skin lesions, but evident performance
disparities between differing skin tones should be addressed before widespread
deployment. In this work, we propose an efficient yet effective algorithm for
automatically labelling the skin tone of lesion images, and use this to
annotate the benchmark ISIC dataset. We subsequently use these automated labels
as the target for two leading bias unlearning techniques towards mitigating
skin tone bias. Our experimental results provide evidence that our skin tone
detection algorithm outperforms existing solutions and that unlearning skin
tone may improve generalisation and can reduce the performance disparity
between melanoma detection in lighter and darker skin tones.

本研究提出了一种高效有效的算法，可以自动标记皮肤损伤图像的肤色，并将其用于注释基准 ISIC 数据集以减少皮肤色调偏差。实验结果证明，皮肤色调检测算法优于现有解决方案，取消偏见可以提高普适性，并减少轻重肤色间的黑色素瘤检测性能差异。