期刊
IEEE SENSORS JOURNAL
卷 19, 期 16, 页码 7107-7119出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JSEN.2019.2913281
关键词
Medical image; images fusion; CNN; MCFNet
资金
- National Natural Science Foundation of China [61501132, 61370084]
- China Postdoctoral Science Foundation [2016M591515]
- Heilongjiang Postdoctoral Scientific Research Developmental Fund [LBH-Q17042]
Medical image fusion technique can help the physician to execute combined diagnosis, preoperative planning, intra-operative guidance, and interventional treatment in many clinical applications by deriving the complementary information from medical images with different modalities. In this paper, we propose an end-to-end deep learning network, multi-layer concatenation fusion network (MCFNet), for feature extraction, feature fusion, and image reconstruction without manually designing complex feature extraction and fusion rules. MCFNet connects the feature map of each layer in the image reconstruction network with the source map, which has the same resolution due to down-sampling. And the spatial information lost in the fused medical image during two down-sampling processes is supplemented. The up-sampling operation at the image reconstruction stage uses fast up-convolution to reduce the checkerboard effect of traditional deconvolution. The proposed fusion method is tested on many clinical medical images and compared with several commonly used image fusion methods. Experimental results show that the proposed fusion method can provide better fusion results in terms of visual quality and objective evaluation.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据