摘要
针对传统的UNet对于大小不一、形状多变的皮肤恶性黑色素瘤图像分割效果不佳的问题,主要通过两点改进实现改进方法对多尺度特征的充分利用,首先在编码器中,采用全局密集网络、局部密集网络以及锯齿状空洞卷积设计,之后在解码器中,采用局部残差设计以及分类正则化。与UNet相比,该改进方法分别在Dice系数、准确率(ACC)、敏感度(SE)、交并比(IOU)指标上提高了0.82%、0.03%、1.99%、1.03%。实验结果证明,改进方法能够提高皮肤恶性黑色素瘤图像分割效果,是一种有效的基础网络结构。
To address the problem that the traditional UNet is ineffective for segmentation of skin malignant melanoma images with variable size and shape, the improved method is mainly implemented to fully utilize the multi-scale features through two improvements, firstly, in the encoder, the global dense network, the local dense network and the dilated convolution design, and later, in the decoder, the local residual design and the classification regularization. Compared with UNet, the improved method improves 0.82%, 0.03%, 1.99%, and 1.03% in the Dice coefficient, accuracy(ACC), sensitivity(SE), and intersection-to-merge ratio(IOU) metrics, respectively. The experimental results demonstrate that the improved method can improve the image segmentation of skin malignant melanoma and is an effective underlying network structure.
作者
赵文慧
杨霄
孟丽洁
Zhao Wenhui;Yang Xiao;Meng Lijie(Military Command System R&D Department,North Automatic Control Technology Institute,Taiyuan 030000,China)
出处
《电子测量技术》
北大核心
2022年第2期110-116,共7页
Electronic Measurement Technology
基金
军委装备发展部预先研究项目(31505550302)资助。
关键词
皮肤病变分割
多尺度特征
空洞卷积
正则化
skin lesion segmentation
multi-scale features
dilated convolution
classification regularization
