论文中稿 IEEE ACCESS 期刊

woodman718

已于 2022-10-09 11:34:44 修改

阅读量1.9k

点赞数 3

分类专栏： Paper 文章标签：深度学习图像处理 pytorch

于 2022-10-04 12:06:07 首次发布

本文链接：https://blog.csdn.net/song5bai/article/details/127160681

版权

Paper 专栏收录该内容

1 篇文章

订阅专栏

提出了一种改进的胶囊网络FixCaps，用于皮肤癌的早期诊断。该模型通过增大底部卷积层核大小并采用注意力模块减少信息损失，相较于现有方法提高了诊断准确性，并减少了大量计算。实验结果显示，在HAM10000数据集上，FixCaps达到了96.49%的准确率。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

奇迹很少发生，生活还需靠自己。

Amazing happens by accident, however, living flourishes with your own strength.

论文：

FixCaps: An Improved Capsules Network for Diagnosis of Skin Cancer | IEEE Journals & Magazine | IEEE Xplore

代码: GitHub - Woodman718/FixCaps: FixCaps: An Improved Capsules Network for Diagnosis of Skin Cancer，DOI: 10.1109/ACCESS.2022.3181225

摘要：

The early detection of skin cancer substantially improves the fifive-year survival rate of patients.

It is often diffificult to distinguish early malignant tumors from skin images, even by expert dermatologists. Therefore, several classifification methods of dermatoscopic images have been proposed, but they have been found to be inadequate or defective for skin cancer detection, and often require a large amount of calculations. This study proposes an improved capsule network called FixCaps for dermoscopic image classifification. FixCaps has a larger receptive fifield than CapsNets by applying a high-performance large-kernel at the bottom convolution layer whose kernel size is as large as 31 × 31, in contrast to commonly used 9 × 9. The convolutional block attention module was used to reduce the losses of spatial information caused by convolution and pooling. The group convolution was used to avoid model underfifitting in the capsule layer. The network can improve the detection accuracy and reduce a great amount of calculations, compared with several existing methods. The experimental results showed that FixCaps is better than IRv2-SA for skin cancer diagnosis, which achieved an accuracy of 96.49% on the HAM10000 dataset.

Generalization Performance

Available:
https://challenge.isic-archive.com/data/#2018
https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/DBW86T

Note: The augmented data of HAM10000 can be obtained as follows: 
https://aistudio.baidu.com/aistudio/datasetdetail/151696

Dataset:  https://www.kaggle.com/datasets/tawsifurrahman/covid19-radiography-database
The COVID-19 Radiography Database consisted of 21165 images.
Among them, covid(3616),normal(10192),opacity(6012),viral(1345).

万物生长，吾需成长：

辗转反侧，徘徊，迷惘，兜兜转转终于在去年十一月初迈进医学图像处理的道路。开始的时候没有多少思路，效果也不理想，后来在跟损友们的交流中找到一丢丢灵感，最后出来的模型效果还不错。当然，期间也跟导师、同侪们多有交流，感谢他们的帮助、支持和鼓励。

IEEE Access 审稿很快：

5月9日投稿

5月11日review

5月19日 Rejected,鼓励修改重投

5月23日大修后重投

6月5日录用

收到录取邮件的时候，激动的心情无以言表。

IEEE access也算SCI检索，中科院3区，还说得过去（满足毕业要求）。

论文的录用，也标志着实验室学习生涯的圆满收官。

感谢蓝老师帮忙报销了版面费，1665美刀，折合人民币一万一，挺贵的。