期刊
IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING
卷 16, 期 4, 页码 854-868出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JSTSP.2022.3181717
关键词
Image fusion; Decoding; Feature extraction; Medical diagnostic imaging; Logic gates; Convolution; Transforms; Information gate module; multimodal medical image fusion; multi-scale cross attention fusion module; saliency weight
资金
- National Natural Science Foundation of China [61966037, 61833005, 61463052, 62066047]
- National Key Research and Development Project of China [2020YFA0714301]
- China Postdoctoral Science Foundation [2017M621586]
- Postgraduate Science Foundation of Yunnan University [2020Z77, CY21624108, 2021Y257, 2021Z45]
- Yunnan Provincial Department of Education Science Foundation [2022Y011]
Multimodal medical image fusion is a technique that aims to merge saliency and complementary information from different source images to assist in biomedical diagnoses. In this paper, the authors propose a new information gate network (IGNFusion) and a Siamese multi-scale cross attention fusion module (SMSCAFM) to optimize the fusion process, achieving significant improvements over existing methods on multiple datasets.
Multimodal medical image fusion aims to merge saliency and complementary information from different source images to assist in biomedical diagnoses. How to effectively utilize feature information in the encoder is a critical issue. However, many existing medical image fusion methods do not consider the contributions of different convolution blocks. In this paper, we propose an information gate module (IGM) to control the contribution of each encoder feature level to the decoder; it is termed the information gate network for multimodal medical image fusion (IGNFusion). Furthermore, the Siamese multi-scale cross attention fusion module (SMSCAFM) integrates saliency and complementary information from multiple source images. Moreover, to constrain the similarity between the fused image and multiple source images, we introduce a saliency weight (SW). Extensive experiments on ten categories of multimodal medical images (i.e., CT & MR-T1 (T1 weighted) and PET & MR-T2 (T2 weighted)) show that our IGNFusion approach achieves significant improvements over 9 state-of-the-art methods.
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