CT图像射束硬化伪影校正方法研究

发布时间：2018-02-10 10:22

本文关键词： 射束硬化校正拉东变换校正模体熵　出处：《沈阳工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：在标准的CT图像重建过程中,被忽视的X射线多能谱特性及其对图像重建算法的影响导致了重建后图像中出现射束硬化伪影,其中包括杯状伪影、条状伪影以及退化的骨边界伪影。这些伪影很大程度上降低了病灶的清晰度,给疾病诊断带来了困扰。同时,在CT机性能评价过程中,图像中的射束硬化伪影会致使测量值偏离实际值,进而影响检测结果。本课题针对射束硬化伪影展开研究,其工作内容包括以下三个方面:首先,针对已有水硬化预校正方法在实际工程应用中的不灵活性,实现了一种以拉东变换为基础的杯状伪影后处理校正方法。使用阈值方法对水模体横断面CT图像进行分割,将分离出的水部分进行拉东变换以获取不同角度下的观测信息;利用该信息进行数学建模并与预先构建好的无杯状伪影模板比较,当两者对比结果的均方误差达到最小时,可计算出所建模型的关键参数。在实际应用中,将临床CT图像软组织的拉东变换结果输入到已求解的校正模型,该模型的输出即为杯状伪影校正后图像。计算机仿真结果表明,所提方法对射束硬化杯状伪影具有较好的抑制效果且对被校正物体的大小及外形无硬性需求。其次,鉴于已有骨硬化伪影校正方法存在校正速度慢或校正过程繁琐的缺点,提出了一种基于校正模体的快速骨伪影校正方法。以人体头部解剖结构为依据构造校正模体,将模体图像中的高衰减物质分割出来并通过单能正投影方法获取相应的投影信息;利用图像的熵与骨硬化伪影存在程度的关系构建数学模型并通过一系列算法确定校正系数;然后平均由5台CT机求解得到的校正系数作为一组通用的校正系数。在实际临床应用中,利用通用校正系数重建伪影图像,将伪影图像叠加在原始图像上得到校正后的CT图像。通过模体和临床数据验证,所提骨硬化伪影校正方法具有校正效果好、校正速度快的优点。所提方法对于被校正图像的卷积核和层厚无要求,具有一定的临床和工程实现价值。最后,将射束硬化校正应用于CT性能指标测量,研究了射束硬化效应对医用全身16排CT机性能测量准确度的影响。实验结果表明,射束硬化对测量平面内空间分辨率的影响与建像参数有关,不同建像参数下的影响趋势基本不同;在层厚测量中,射束硬化会致使测量值大于实际值;而在低对比度分辨率测量方面,射束硬化对其影响甚微。
[Abstract]:In the standard CT image reconstruction process, the neglected X-ray multispectral characteristics and its influence on the image reconstruction algorithm lead to the emergence of beam hardening artifacts, including goblet artifacts, in the reconstructed images. Strip artifacts and degenerated bone boundary artifacts. These artifacts greatly reduce the focus's clarity and cause trouble for disease diagnosis. At the same time, during the evaluation of CT performance, The beam hardening artifacts in the image will cause the measured values to deviate from the actual values, and then affect the detection results. In this paper, the beam hardening artifacts are studied. The work includes the following three aspects: first, In view of the inflexibility of existing water hardening precorrection methods in practical engineering applications, a new method of post-processing of cup pseudo-image based on Radon transform is implemented. The threshold method is used to segment the cross-sectional CT images of water motifs. The separated water part is transformed to obtain the observation information at different angles. The information is used for mathematical modeling and compared with the pre-constructed non-goblet artifact template, when the mean square error of the two comparison results reaches the minimum. The key parameters of the model can be calculated. In practical application, the results of Rato transform of soft tissue of clinical CT images are input to the corrected model, and the output of the model is the image corrected by goblet artifacts. The computer simulation results show that, The proposed method has good suppression effect on beam hardening goblet artifacts and has no rigid requirements for the size and shape of the corrected objects. Secondly, because of the shortcomings of the existing methods for correction of bone hardening artifacts, the correction speed is slow or the correction process is cumbersome. A fast bone artifact correction method based on correction motif is proposed. Based on the anatomical structure of human head, the high attenuation material in the motif image is segmented and the corresponding projection information is obtained by single energy orthographic projection. Based on the relationship between image entropy and the degree of existence of bone sclerosis artifact, a mathematical model is constructed and the correction coefficient is determined by a series of algorithms. Then the average correction coefficients obtained by 5 CT machines are taken as a set of general correction coefficients. In practical clinical applications, the artifacts are reconstructed by using the general correction coefficients. The corrected CT image was obtained by superposing the artifact image on the original image. Through the verification of the motif and clinical data, the correction method of bone sclerosis artifact was proved to be effective. The proposed method has no requirement for convolution kernel and slice thickness of corrected image, and has certain clinical and engineering value. Finally, beam hardening correction is applied to the measurement of CT performance index. The effect of beam hardening on the measurement accuracy of medical 16-slice CT machine is studied. The experimental results show that the effect of beam hardening on the spatial resolution in the measurement plane is related to the imaging parameters. In thickness measurement, beam hardening will cause the measured value to be larger than the actual value, but in low contrast resolution measurement, beam hardening has little effect on it.
【学位授予单位】：沈阳工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】