优化平稳小波变换去除功能MRI数据噪声的方法

Optimizing stationary wavelet transform denoising of functional MRI data

目的通过优化小波变换多分辨率去噪方法,在保持一定灵敏度的条件下,降低假阳性率。方法在小波重构时较原来的方法保留更多的小波尺度,对原有方法的分析结果进行多人平均,并用模拟数据和视觉实验数据对这些方法进行验证。结果分析模拟数据显示,在α<0.01条件下,本文方法能在保持一定灵敏度的基础上有效地克服原有方法假阳性率高的缺点。分析实验数据显示,以SPM2为标准,在α<0.001条件下,本文方法能给出既灵敏又相对准确的结果。结论本文方法能同时兼顾灵敏度和准确度,是原有方法的一种优化。

Objective To optimize the original denoising method of wavelet multiresolution analysis in order to decrease false positive rate while keeping the certain sensitivity. Methods More wavelet decomposition scales are chosen when wavelet reconstructing, and strategy of multi-subjects averaging is conducted. These proposed methods are validated with simulated and visual experimental data. Results Analyzing simulated data reveals that,when a〈0. 01, the proposed methods can, with the certain sensitivity, efficiently alleviate the defect of the original method. Analyzing the experimental data reveals that, judged by the criterion of SPM2, when a〈0. 01, the proposed method can give both rather accurate and highly sensitive result. Conclusion The proposed method is the optimization of the original denoising method,making the better compromise between sensitivity and accuracy