| 支联合,李玉晓,赵书俊,单保慈.基于离散小波变换的fMRI数据特征提取[J].中国医学影像技术,2010,26(6):1151~1154 |
| 基于离散小波变换的fMRI数据特征提取 |
| Feature extraction of fMRI data based on discrete wavelet transform |
| 投稿时间：2009-11-23 修订日期：2010-03-06 |
| DOI： |
| 中文关键词: 磁共振成像 特征提取 离散小波变换 相关分析 |
| 英文关键词:Magnetic resonance imaging Feature extraction Discrete wavelet transform Correlation analysis |
| 基金项目:周口师院博士科研启动基金(2006SRFD002)、河南省教育厅自然科学研究计划项目(2007310024、2008A180040)。 |
|
| 摘要点击次数: 4376 |
| 全文下载次数: 915 |
| 中文摘要: |
| 目的 设计一种灵敏度高且处理速度快的fMRI数据小波分析方法。方法 先用离散小波变换和频谱分析确定有用信号存在的小波分解尺度,也即特征尺度;再对实验数据进行离散小波分解,重构时将非特征尺度里的小波系数设置为0;再以相关分析对小波重构信号进行激活检测。结果 对视觉数据的分析结果显示,新方法的灵敏度与基于平稳小波变换、SPM2方法相当,而优于基于提升小波变换的方法;新方法的处理速度与基于提升小波变换的方法相当,但较平稳小波变换方法有较大提高。结论 本文为fMRI数据提供了一种更为灵敏且快速的小波分析方法,更为实用。 |
| 英文摘要: |
| Objective To design a sensitive and fast wavelet-based analysis approach for fMRI data. Methods The wavelet scales in which the useful signal exists, feature scales termed in the paper, were first discerned with discrete wavelet transform (DWT) and frequency analysis. Then fMRI data were decomposed with DWT into wavelet coefficients at different wavelet scales, but those at non-feature scales were set to zero while conducting the discrete wavelet reconstruction. Finally the reconstructed signals were subjected to correlation analysis for detecting active pixels. Results Analyzing the visual fMRI data indicated that the sensitivity of the proposed approach was the same as the one based on the stationary wavelet transform and SPM2, but superior to the one based on the lifting wavelet transform. The process speed of the proposed approach was near to the one based on the lifting wavelet transform, but much faster than the one based on the stationary wavelet transform. Conclusion The method provides a sensitive and fast wavelet-based analysis approach for fMRI data. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
