Development of a Reconstruction Method using the Non-uniform Fourier Transform and a Machine Learning Approach for Spiral Imaging [Presidential Award Proceedings]
-
- TAMADA Daiki
- Department of Radiology, University of Yamanashi
-
- KOSE Ryoichi
- MRIsimulations Inc.
-
- KOSE Katsumi
- MRIsimulations Inc.
-
- WAKAYAMA Tetsuya
- GE Healthcare Japan
-
- ONISHI Hiroshi
- Department of Radiology, University of Yamanashi
-
- MOTOSUGI Utaroh
- Department of Radiology, University of Yamanashi
Bibliographic Information
- Other Title
-
- 不等間隔高速フーリエ変換と機械学習を用いたSpiral再構成手法の開発[大会長賞記録]
Search this article
Abstract
<p> Spiral magnetic resonance imaging enables fast and efficient acquisition, although it suffers from aliasing artifact in cases of undersampling. It is still challenging to achieve fast and robust reconstruction for undersampled spiral datasets since iterative approaches are generally required. In this study, we developed a straightforward reconstruction method for spiral imaging using the non-uniform Fourier transform and convolutional neural networks (CNN). Two CNNs were used for estimating the sensitivity maps for parallel imaging and removing aliasing artifact caused by the undersampling. Simulation and volunteer studies were conducted to show the usefulness of our method. The results implied that the proposed method enables better performance than a conventional reconstruction method.</p>
Journal
-
- Japanese Journal of Magnetic Resonance in Medicine
-
Japanese Journal of Magnetic Resonance in Medicine 39 (1), 20-24, 2019-02-15
Japanese Society for Magnetic Resonance in Medicine
- Tweet
Keywords
Details 詳細情報について
-
- CRID
- 1390845713057077376
-
- NII Article ID
- 130007615384
-
- ISSN
- 24340499
- 09149457
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
- Crossref
- CiNii Articles
-
- Abstract License Flag
- Disallowed