摘要肺4D-CT在肺癌放射治疗中发挥着重要的作用，但肺4D-CT数据层间的分辨率低，导致每个相位3D数据的肺冠矢状面均为低分辨率图像。本研究提出一种基于运动估计的超分辨率重建技术，以提高3D数据的冠矢状面图像分辨率。首先，分析图像退化模型；然后，采用基于完全搜索块匹配的运动估计法，估计出不同“帧”肺冠矢状面图像之间的运动场；最后，以此运动场为基础，采用迭代反投影法（IBP），重建高分辨率的肺部冠矢状面图像。使用一个公共可用的数据集来评价所提出的算法，该数据集由10组肺4D-CT数据组成，每组数据包含10个相位。在每组图像中，选取不同相位的冠矢状面图像进行实验。结果表明，与传统的插值方法（如最近邻插值、双线性插值法）相比，图像边缘宽度均显著降低（最近邻插值9.93±0.59，双线性插值8.04±0.69，新算法5.41±0.60， P<0.001）；较双线性插值，图像平均梯度显著提高（5.41±0.59 vs 7.49±0.75, P<0.001），新方法不仅能获得视觉上清晰的图像，而且量化评价指标也有明显提高。主观和客观实验结果表明，所提出的新方法能有效提高肺4D-CT冠矢状面图像的分辨率。
Abstract：Lung 4D-CT plays an important role in the lung cancer radiotherapy. However, due to the great interslice thickness, the coronal and sagittal images of each 3D phase data are with low resolution. In this paper, we proposed a superresolution reconstruction method based on motion estimation to improve the image resolution of lung coronal and sagittal images. First, we analyzed the basic model of image degradation. Then, we employed motion estimation method based on full search block matching algorithm to estimate images motion vector field of different “frames”. At last, based on the motion vector field, we employed the iterative back projection (IBP) algorithm to reconstruct highresolution lung coronal and sagittal images. We used a public available dataset to evaluate the proposed algorithm, the dataset consists of 10 groups of lung 4D-CT data, each group contains 10 phases. We selected sagittal images of different phases to perform above procedure for experiments. Experimental results showed that the proposed algorithm significantly decreases edge widths（5.41±0.60, P<0.001）compared with the conventional methods, for instance, the nearest interpolation is 9.93±0.59 and bilinear interpolation is 8.04SymbolqB@069, which significantly increases image average gradient value compared with bilinear interpolation (7.49±0.75 vs 5.41±0.59, P<0.001）. Our method yields superior performance both qualitatively and quantitatively.
［1］Khan F, Bell G, Antony J, et al. The use of 4DCT to reduce lung dose: A dosimetric analysis［J］. Medical Dosimetry, 2010, 34(4): 273-278.
［2］Li T, Schreibmann E, Thorndyke B, et al. Radiation dose reduction in fourdimensional computed tomography［J］. Medical Physics, 2005,32(12):3650-3660.
［3］浦剑, 张军平, 黄华. 超分辨率算法研究综述［J］. 山东大学学报 (工学版), 2009, 39(1): 27-32.
［4］De Vos L, Stegherr M. Parameterizable VLSI architectures for the fullsearch blockmatching algorithm［J］. IEEE Transactions on Circuits and Systems, 1989, 36(10): 1309-1316.
［5］郭伟伟, 章品正. 基于迭代反投影的超分辨率图像重建［J］. 计算机科学与探索, 2009, 3(3): 321-329.
［6］李桐. 超分辨率图像重建技术［J］. 哈尔滨师范大学自然科学学报, 2006, 5: 69-71.
［7］Rousseau F. A nonlocal approach for image superresolution using intermodality priors ［J］. Medical Image Analysis, 2010, 14(4): 594-605.
［8］黄新生, 杨庆伟, 王亦平, 等. 图像序列运动估计技术综述［J］. 计算机仿真, 2008, 5:180-183.
［9］朱长征, 沈振康. 一种改进的完全搜索块匹配算法［J］. 红外与激光工程, 2004, 33(4): 389-391.
［10］Irani M, Peleg S. Motion analysis for image enhancement: Resolution, occlusion, and transparency ［J］. Journal of Visual Communication and Image Representation, 1993, 4(4): 324-335.
［11］Castillo R, Castillo E, Guerra R, et al. A framework for evaluation of deformable image registration spatial accuracy using large landmark point sets［J］. Physics in Medicine and Biology, 2009, 54(7): 1849-1870.
［12］Greenspan H, Oz G, Kiryati N, et al. MRI interslice reconstruction using superresolution［J］. Magnetic Resonance Imaging, 2002, 20(5): 437-446.［13］徐美芳, 刘晶红. 基于边缘保持的航拍图像凸集投影超分辨率重建算法［J］. 液晶与显示, 2010, 25(6): 873-878.
［14］Chan SH, V DT, Nguyen TQ. Subpixel motion estimation without interpolation［C］// Acoustics Speech and Signal Processing (ICASSP), Dallas: IEEE, 2010: 722-725.
