北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2022, Vol. 45 ›› Issue (5): 97-102.

• 研究报告 • 上一篇    下一篇

基于稀疏变换学习的改进灵敏度编码重建算法

李玺兰,段继忠   

  1. 昆明理工大学 信息工程与自动化学院
  • 收稿日期:2021-10-20 修回日期:2021-12-05 出版日期:2022-10-28 发布日期:2022-11-01
  • 通讯作者: 段继忠 E-mail:duanjz@kust.edu.cn
  • 基金资助:
    国家自然科学基金项目

An Improved Sensitivity Encoding Reconstruction Algorithm Based on Sparse Transform Learning

LI Xilan, DUAN Jizhong   

  • Received:2021-10-20 Revised:2021-12-05 Online:2022-10-28 Published:2022-11-01
  • Contact: DUAN Jizhong E-mail:duanjz@kust.edu.cn

摘要: 灵敏度编码(SENSE)是一种利用多个接收线圈的灵敏度信息来减少扫描时间的技术基于 SENSE 算法的磁共振成像重建算法的重建图像存在模糊伪影和细节缺失等问题,不利于临床医学诊断为减少磁共振重建图像伪影并提高重建图像质量,将数据驱动的自适应稀疏变换学习(TL)引入 SENSE 算法中,得到一种 TL-SENSE 算法该算法利用交替方向乘子法进行求解,通过变换更新硬阈值去噪和图像更新实现并行磁共振成像重建仿真实验结果表明,所提算法对图像去噪和修复效果较好,能完整保留纹理细节和边缘轮廓信息,目标图像与原始图像的一致性较高对所选 48 组数据, TL-SENSE 算法重建图像的平均信噪比相比 SENSE 算法、 L1-SENSE 算法、TV-SENSE算法和 LpTV-SENSE 算法,分别提高了 4.62 dB、1.91 dB、1.30 dB 0.89 dB。

关键词: 并行磁共振成像, 变换学习, 灵敏度编码, 交替方向乘子法

Abstract:

The sensitivity encoding ( SENSE) technique utilizes sensitivity information from multiple receiving coils to reduce scan time. The existing SENSE-based parallel MRI reconstruction methods have problems of artifacts and missing details, which is not conducive to clinical diagnosis. By introducing data-driven adaptive sparse transform learning (TL) into the SENSE algorithm, TL-SENSE algorithm is proposed, that reduce the artifacts and improve the quality of parallel MRI reconstruction. The proposed algorithm employs the alternating direction method of multipliers (ADMM) to solve the target optimization problem. And The proposed algorithm comprises three steps: transform updating, hard threshold denoising and image updating. The simulation results show that the proposed algorithm performs well in image denoising and restoration and preserves the texture details and edge information. It also achieves higher consistency between the reconstructed image and the original image. For the selected 48 sets of data, the average signal noise ratio of TL-SENSE increased by 4.62 dB, 1.91 dB, 1.30 dB and 0.89 dB compared with that of SENSE, L1-SENSE, TV-SENSE, and LpTV-SENSE, respectively.

Key words: parallel magnetic resonance imaging, transform learning, sensitivity encoding, alternating direction method of multipliers

中图分类号: