【作者】 谢丹；

【导师】 徐晓嵘； 田超；

【作者基本信息】 中国科学技术大学 ， 仪器科学与技术， 2023， 硕士

【摘要】 基于光声效应发展起来的光声断层成像技术(Photoacoustic Computed Tomography,PACT)能够对从细胞器到器官等不同尺度的生物结构进行高分辨率成像。然而在实际情况中,PACT成像受到换能器阵元口径、带宽、探测视角等因素的影响,重建图像往往会出现不同程度的模糊。考虑到PACT成像系统的点扩散函数(Point Spread Function,PSF)通常是空间变化的(以下简称空变),本文研究提出了一种基于空变解卷积的光声图像像质提升方法。首先,本文研究了 PACT系统空变PSF的获取方法。搭建了一套精密实验装置对系统中部分预设位置处的PSF进行了测量,然后利用主成分分析拟合上述PSF,将其转换为一系列基函数与对应权重的线性加权和,从而构建了空变PSF表征模型;接着,再利用径向基函数插值法获取未采样点处PSF的权重,基于上述表征模型和插值参数获得像面完整的PSF。其次,本文在贝叶斯最大后验估计框架下对空变模糊的光声图像复原问题进行了建模。基于系统测量误差服从泊松分布的假设,推导出了适用于空变PSF成像系统的Richardson-Lucy复原算法(以下简称RL算法);为了抑制复原过程中的噪声,研究提出了稀疏对数梯度(Sparse Logarithmic Gradient,SLG)正则化约束方法,并对RL算法进行了改进(以下简称SLG-RL算法),提升复原过程的稳定性。接着,本文基于常用的环形阵列光声换能器的成像特点,提出了一种新的去伪影方法。该方法根据条纹伪影的周期性分布规律,通过图像的旋转叠加操作并结合SLG-RL算法实施逆向解算,可有效抑制光声图像中的条纹伪影并保留图像细节。最后,本文开展了仿真、仿体、活体实验以验证上述方法的准确性,所有结果均表明本文提出的方法能有效去除光声图像中的空变模糊和条纹伪影,显著改善图像像质。此外,本文还比较了不同正则项的复原效果,复原结果表明,本文所提出的SLG正则项性能优于传统的Tikhonov正则项,能得到更好的复原图像。综上所述,本文提出的像质提升方法对于更好地发挥PACT成像技术在医疗诊断中的作用具有重要意义。更多还原

【Abstract】 Photoacoustic computed tomography(PACT)is a novel,noninvasive biomedical imaging modality and can visualize biological structures at multiply scales ranging from organelles to organs.However,due to the non-ideal imaging condition like finite detector aperture size,bandwidth or limited view angle,the reconstructed PACT images usually suffer from spatially variant blur,which severely damage the image details.Therefore,we propose a restoration method which is based on deconvolution with spatially variant point spread function(PSF)to improve the quality of the PACT images.Firstly,we design a precise measurement method to obtain the PSFs of certain discrete positions in the PACT system,then we use the principal component analysis(PCA)to transform the obtained spatially variant PSF into a linear combination of some spatially invariant basis PSFs.Therefore,the representation model of spatially variant PSF is constructed.Next,the radial basis interpolation is adopted to achieve the entire spatially variant PSF of the PACT system.Secondly,we describe the problem of image restoration under the framework of Bayesian maximum a posteriori probability.Based on the assumption that the measurement error of the PACT system follows Poisson distribution,we derive the Richardson-Lucy(RL)algorithm with spatially variant PSF.In order to suppress the noise during the image restoration,a new regularization method named sparse logarithmic gradient(SLG)regularization is proposed.We integrate SLG with the RL algorithm(SLG-RL)to improve the stability of restoration.Thirdly,we design a method to remove the streak artifacts in the PACT images obtained with a ring-shape transducer array.Considering the imaging characteristics of ring-shape transducer array,the streak artifacts oscillate approximately periodically,so this method firstly adopts the rotation and average operations to suppress the streak artifacts,then the SLG-RL is adopted once more to remove the blur resulting from rotation and average operations.Finally,we carry out simulation,phantom and in vivo experiments.All the results show that our restoration method can improve the image resolution and remove the streak artifacts effectively.In addition,we also compare the restoration results with different regularizations.The results show that the SLG regularization outperforms classic Tikhonov regularization.In conclusion,the method we proposed is of great significance to better play the role of PACT imaging technology in medical diagnosis.更多还原