【作者】 尹愚；

【导师】 尧德中；

【作者基本信息】 电子科技大学 ， 生物医学工程， 2011， 博士

【摘要】 脑内源信号光学成像是近十几年发展起来的研究大脑皮层神经群体活动的成像方法。该方法具有较高的空间分辨率,主要应用于大脑皮层功能构筑的研究。脑内源信号光学成像是一种二次成像技术,其信号非常微弱并带有皮层生理结构所决定的信号特征。有效的空间滤波方式能对图像噪声进行抑制,使微弱信号能得以增强。本文针对脑内源信号光学成像,对所采集的图像结果使用不同的线性与非线性空间滤波处理,对不同滤波方法的噪声抑制程度和图像细节保存度进行讨论。结果显示非线性中值空域滤波对内源性光学图像的去噪效果明显,并能保留清晰的信号细节特征,是适于脑内源性光学成像信号特点的图像增强技术。通过对脑内源性光学信号噪声来源和皮层功能柱生理结构特征的讨论,对标准化处理后的功能图像进行二阶统计量分析。根据信号的空间结构特性提出了自适应的滤波窗口设计,对含有不同特征的区域分别采用不同的滤波窗口进行中值空域滤波处理。获得的图像结果去除噪声效果良好,并充分的保留了图像功能信号的细节。内源性光学信号不仅微弱,而且在穿过皮层组织时会发生扩散,使得图像存在一定程度的退化。通过选取功能柱区域与非功能柱区域像素进行联合求解高斯函数方程,得到内源性光学信号的点扩展函数的估计,并对内源性光学信号图像进行恢复。恢复图像的二阶统计量的分布形式表明,图像中功能柱区域的范围较处理前更为集中。利用此结果,以恢复后图像的二阶统计量作为阈值划分标准,对图像准确的分割提取出功能柱区域。经过去噪和反卷积恢复的功能图像,通过选取激活区域和非激活区域的像素分别进行时间过程对比分析,在结果中观察到一种全新的快速功能成份。通过对内源信号成像过程的血氧动力学的讨论,该快速成份可能表征的是皮层间相互抑制作用。更多还原

【Abstract】 Optical imaging based on intrinsic signals is a newly emerged technology with highest spatial resolution among in vivo brain research. It is a useful technology for studying functional architecture of the cortex in a large scale. Optical imaging of intrinsic signals is a secondary image of the cerebral cortex. The weak optical signal is decided by anatomical structure of brain.The spatial filter is a powerful technology for de-noising and image enhancement. We used different linear filter and nonlinear to deal with optical imaging. Furthermore, we compared the degree of noise suppression and image details. Our result shows that nonlinear median filter can keep more image details with effective noise reduction. It is useful for image enhancement of optical imaging.For effectively extracting the endogenous weak optical signal, we analyzed the sources of noise and the signal characteristics in view of the anatomical structures of the brain. The present study explored the second-order statistical results of function image after preprocessing and presented an adaptive filter windows design. Our result shows that this algorithm can keep more image details with effective noise reduction.The present study explored the second-order statistical results of function image,and proposed an estimation method for the point spread function of endogenous optical signals based on select different pixels to solute Gauss equations. The second-order statistics of deconvolution results showed that this algorithm keep more details with function columns. The classification method for the threshold extracted from second-order statistical results of deconvolution image was proposed to ensure the accurate division of the activated functional orientation columns.Temporal analysis was used to different parts of function image that have been de-noised and deconvoluted. In timecourse curve, we found a new fast functional component. This fast component may show the inhibition of cortex, according to discussion of the results and hemodynamic of intrinsic signal.更多还原