光子计数能谱CT重建算法与系统优化研究

发布时间：2018-05-25 21:29

本文选题：能谱CT + 光子计数探测器　；参考：《清华大学》2016年博士论文

【摘要】：CT是现代医学必不可少的检查手段,在许多疾病的诊断中都起着不可替代的作用。随着光子计数探测器技术的发展,光子计数能谱CT得到了越来越多的关注。它可以提供传统CT所没有的物质分辨能力,对造影成像和软组织成像具有重要意义;同时它比双能CT噪声更低,可以降低患者所受的剂量。本文以降低能谱CT重建图像噪声为目标,从重建算法和探测器系统优化两方面对能谱CT进行了研究。在重建算法方面,针对投影数据信息冗余的特点,本文提出了一种基于加权子集的投影分解算法。它将能区拆分成不同的组合,对每一个组合使用多项式进行物质分解。然后利用标定数据计算权重系数对它们进行加权组合,从而极小化分解噪声。模拟和实验结果表明所提出的方法接近可能达到的最小分解噪声。物质分解图像重建过程中,我们利用物质分解图像与衰减图像的相似性,提出了加权非局部TV极小化的优化重建算法。通过计算衰减图像像素之间的相似性作为权重,构造加权非局部TV的目标函数并使用ASD-POCS算法求解,并对算法主要参数的影响进行了讨论并制订了自适应选择策略。模拟和实验结果表明所提出的算法相比于一些已有方法边缘保持性能更好。探测器系统优化的重点是阈值优化。我们首先对探测器的能量响应函数进行了建模,以求解阈值优化所需的等效能谱。本文提出了一种混合蒙特卡洛模型,使用GEANT4模拟X射线光子进入探测器的过程,然后使用简化数值模型描述后续的统计涨落和电荷共享。模型中加入了探测器的抗电荷共享逻辑。进一步通过测量金属荧光谱拟合出模型的关键参数,并使用光机能谱对能量响应函数模型的有效性进行了验证。我们进一步提出了投影概率密度函数加权的CRLB作为探测器阈值优化目标函数。结合探测器的能量响应函数,使用差分进化算法对探测器阈值进行优化以极小化加权CRLB,从而实现对重建图像噪声水平的极小化。模拟和实验结果验证了算法的有效性和对参数选择的鲁棒性。基于这个方法,进一步研究了探测器主要参数对重建噪声和最优阈值的影响规律。
[Abstract]:Ct is an indispensable examination method in modern medicine and plays an irreplaceable role in the diagnosis of many diseases. With the development of photon counting detector technology, photon counting spectrum CT has been paid more and more attention. It can provide material resolution that traditional CT does not have, and has important significance for imaging and soft tissue imaging. At the same time, it has lower noise than dual-energy CT and can reduce the dose of patients. The aim of this paper is to reduce the noise of the reconstructed image with the aim of reducing the noise of the reconstructed image. The spectral CT is studied from two aspects: the reconstruction algorithm and the system optimization of the detector. In the aspect of reconstruction algorithm, a projection decomposition algorithm based on weighted subset is proposed in this paper, according to the characteristics of information redundancy of projection data. It splits the energy region into different combinations and decomposes each of them using polynomials. Then the weighted coefficients are calculated by the calibration data to minimize the decomposition noise. The simulation and experimental results show that the proposed method is close to the minimum decomposition noise that can be achieved. In the process of material decomposition image reconstruction, we propose a weighted nonlocal TV minimization algorithm based on the similarity between the material decomposition image and the attenuated image. By calculating the similarity between attenuated image pixels as weights, the objective function of weighted non-local TV is constructed and solved by ASD-POCS algorithm. The influence of the main parameters of the algorithm is discussed and an adaptive selection strategy is worked out. Simulation and experimental results show that the proposed algorithm has better edge retention performance than some existing methods. The focus of detector system optimization is threshold optimization. The energy response function of the detector is first modeled to solve the equivalent energy spectrum required for threshold optimization. In this paper, a hybrid Monte Carlo model is proposed, in which GEANT4 is used to simulate the process of X-ray photons entering the detector, and then a simplified numerical model is used to describe the subsequent statistical fluctuations and charge sharing. The anti-charge sharing logic of the detector is added to the model. Furthermore, the key parameters of the model are fitted by measuring the metal fluorescence spectrum, and the validity of the energy response function model is verified by optical spectroscopy. Furthermore, we propose CRLB weighted by projection probability density function as the objective function of detector threshold optimization. Combined with the energy response function of the detector, the threshold of the detector is optimized by using the differential evolution algorithm to minimize the weighted CRLBs, so that the noise level of the reconstructed image can be minimized. Simulation and experimental results show that the algorithm is effective and robust to parameter selection. Based on this method, the influence of the main parameters of the detector on the reconstruction noise and the optimal threshold is further studied.
【学位授予单位】：清华大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP391.41;R814.42

【相似文献】