卷积神经网络
计算机科学
人工智能
降噪
深度学习
学习迁移
噪音(视频)
模式识别(心理学)
还原(数学)
特征(语言学)
词典学习
计算机视觉
图像(数学)
数学
语言学
哲学
几何学
作者
Rongbiao Yan,Yi Liu,Yuhang Liu,Lei Wang,Rongge Zhao,Yunjiao Bai,Zhiguo Gui
出处
期刊:IEEE transactions on computational imaging
日期:2023-01-01
卷期号:9: 83-93
被引量:11
标识
DOI:10.1109/tci.2023.3241546
摘要
Removing noise and artifacts from low-dose computed tomography (LDCT) is a challenging task, and most existing image-based algorithms tend to blur the results. To improve the resolution of denoising results, we combine convolutional dictionary learning and convolutional neural network (CNN), and propose a transfer learning densely connected convolutional dictionary learning (TLD-CDL) framework. In detail, we first introduce the dense connections and multi-scale Inception structure to the network, and train the pre-model on the natural image dataset, then fit the model to the post-processing of LDCT images in the way of transfer learning. In addition, considering that a single pixel-level loss is difficult to achieve satisfactory results both in the index and visual perception, we use the compound loss function of L1 loss and SSIM loss to guide the training. The experimental result shows that TLD-CDL has a good balance between noise reduction and the preservation of details, and acquires inspiring effectiveness in terms of qualitative and quantitative perspective.
科研通智能强力驱动
Strongly Powered by AbleSci AI