Ring artifact is the major factor that seriously influence quality of computed tomography(CT)image reconstruction,especially in testing large-scale workpieces.To remove ring artifact quickly and efficiently,a kind of ...Ring artifact is the major factor that seriously influence quality of computed tomography(CT)image reconstruction,especially in testing large-scale workpieces.To remove ring artifact quickly and efficiently,a kind of ring artifact correction method is improved based on the post-processing CT image reconstruction in this paper.At first,transform the CT image from the rectangular coordinates into polar coordinates.Next,design multidimensional filter to filter the image and calculate the mean and variance of each filtered pixel in polar coordinates.The location of artifact point is determined by the double precision.One is the comparison of calculated variance and variance threshold,and another is the comparison of pixel value and pixel value threshold.Then,process the artifact points in a reasonable manner and do details remain to CT image in particular.At last,convert corrected polar image into rectangular coordinates.The actual experiment shows that compared with the original method,improved method can better correct the ring artifacts and keep the image details for CT images.It is a kind of practical ring artifact correction methods for subsequent processing and quantitative analysis.展开更多
乳腺肿瘤是一种常见的恶性肿瘤,其临床诊断不但费时费力,还容易出现误诊.本文旨在建立一个基于乳腺数据自动分割的乳腺肿瘤计算机辅助诊断模型,提高临床诊断的速度和准确率.为了用卷积神经网络U-Net模型分割对比增强锥光束乳腺计算机断...乳腺肿瘤是一种常见的恶性肿瘤,其临床诊断不但费时费力,还容易出现误诊.本文旨在建立一个基于乳腺数据自动分割的乳腺肿瘤计算机辅助诊断模型,提高临床诊断的速度和准确率.为了用卷积神经网络U-Net模型分割对比增强锥光束乳腺计算机断层扫描(Contrast-Enhanced Cone-Beam Breast CT,CE-CBBCT)数据,本文首先沿冠状面将3维数据转换成2维切片,通过数据默认的窗口对其进行归一化处理.实验结果显示,使用U-Net卷积神经网络对数据进行分割,Dice系数和IoU(Intersection over Union)分别为0.7920和0.6962.然后,本文用不同骨干网络(即各种深度学习分类网络)去替换U-Net的编码器,再次进行分割并对比不同特征提取对分割性能的影响,发现旋转增广方式可以提升各分割网络的性能.其中,基于ResNet152的U形分割网络的性能最好,Dice系数和IoU分别达到0.8410和0.7576.随后,本文又在所有模型中选取5个性能最好的模型组成一个集成模型,重复分割实验,发现此模型有最佳分割性能,平均Dice系数和IoU可达0.8463和0.7676,性能显著提升.值得指出的是,在处理数据时本文仅使用数据默认的窗口,降低了对人工的依赖.展开更多
Aim To investigate the relationship between the positioning of the lower central incisor and physical morphology of the surrounding alveolar bone. Methodology Thirty-eight patients （18 males, 20 females）, with mean ...Aim To investigate the relationship between the positioning of the lower central incisor and physical morphology of the surrounding alveolar bone. Methodology Thirty-eight patients （18 males, 20 females）, with mean age of 13.4 years, were included in this study. As part of orthodontic treatment planning the patients were required to take dental Cone-beam CT （CBCT） covering the region of lower incisors, the sur- rounding alveolar bone and the mandibular symphysis. The cephalometric parameters were designed and measured to indicate the inclination of lower central incisor and physical morphology of the adjacent alveolar bone. Computer-aided descriptive statistical analysis was performed using SPSS 15.0 software package for Windows. A correlation analysis and a linear regression analysis between the incisor inclination and the alveolar bone morphology were performed. Results Significant positive correlations were found between the lower central incisor inclination and the morphological contour of the alveolar bone （P〈0.05）. The lower central incisor root apex was closer to the lingual alveolar crest when it was buccally inclined. Conclusion The morphology of the alveolar bone may be affected by incisal inclination.展开更多
基于实验室设备自主开发了一套320 kV锥束工业CT成像系统。首先设计了基于运动控制卡的六轴机械扫描平台,并将其与320 kV射线机、平板探测器结合搭建了锥束CT成像系统的硬件。其次,利用平板探测器和运动控制卡的动态链接库开发锥束CT投...基于实验室设备自主开发了一套320 kV锥束工业CT成像系统。首先设计了基于运动控制卡的六轴机械扫描平台,并将其与320 kV射线机、平板探测器结合搭建了锥束CT成像系统的硬件。其次,利用平板探测器和运动控制卡的动态链接库开发锥束CT投影数据采集软件和重建软件,自主开发FDK算法并基于CUDA架构实现FDK算法的GPU加速。最后,采用阻尼器对锥束CT成像系统进行测试。结果表明,整个系统运行稳定,可以获得高质量的CT重建图像,且重建效果与VG STUDIO MAX 3.4软件的重建效果一致。展开更多
Because of the growing concern over the radiation dose delivered to patients, X-ray cone-beam CT(CBCT) imaging of low dose is of great interest. It is difficult for traditional reconstruction methods such as Feldkamp ...Because of the growing concern over the radiation dose delivered to patients, X-ray cone-beam CT(CBCT) imaging of low dose is of great interest. It is difficult for traditional reconstruction methods such as Feldkamp to reduce noise and keep resolution at low doses. A typical method to solve this problem is using optimizationbased methods with careful modeling of physics and additional constraints. However, it is computationally expensive and very time-consuming to reach an optimal solution. Recently, some pioneering work applying deep neural networks had some success in characterizing and removing artifacts from a low-dose data set. In this study,we incorporate imaging physics for a cone-beam CT into a residual convolutional neural network and propose a new end-to-end deep learning-based method for slice-wise reconstruction. By transferring 3D projection to a 2D problem with a noise reduction property, we can not only obtain reconstructions of high image quality, but also lower the computational complexity. The proposed network is composed of three serially connected sub-networks: a cone-to-fan transformation sub-network, a 2D analytical inversion sub-network, and an image refinement sub-network. This provides a comprehensive solution for end-to-end reconstruction for CBCT. The advantages of our method are that the network can simplify a 3D reconstruction problem to a 2D slice-wise reconstruction problem and can complete reconstruction in an end-to-end manner with the system matrix integrated into the network design. Furthermore, reconstruction can be less computationally expensive and easily parallelizable compared with iterative reconstruction methods.展开更多
The accuracy of conventional superposition or convolution methods for scatter correction in kV-CBCT is usually compromised by the spatial variation of pencil-beam scatter kernel (PBSK) due to finite size, irregular ex...The accuracy of conventional superposition or convolution methods for scatter correction in kV-CBCT is usually compromised by the spatial variation of pencil-beam scatter kernel (PBSK) due to finite size, irregular external contour and heterogeneity of the imaged object. This study aims to propose an analytical method to quantify the Compton single scatter (CSS) component of the PBSK, which dominates the spatial distribution of total scatter assuming that multiple scatter can be estimated as a constant background and Rayleigh scatter is the secondary source of scatter. The CSS component of PBSK is the line integration of scatter production by incident primary photons along the beam line followed by the post-scattering attenuation as the scattered photons traverse the object. We propose to separate the object-specific attenuation term from the line integration and equivalently replace it with an average value such that the line integration of scatter production is object independent but only beam specific. We derived a quartic function formula as an approximate solution to the spatial distribution of the unattenuated CSS component of PBSK. The “effective scattering center” is introduced to calculate the average attenuation. The proposed analytical framework to calculate the CSS was evaluated using parameter settings of the On-Board Imager kV-CBCT system and was found to be in high agreement with the reference results. The proposed method shows highly increased computational efficiency compared to conventional analytical calculation methods based on point scattering model. It is also potentially useful for correcting the spatial variant PBSK in adaptive superposition method.展开更多
基金National Natural Science Foundation of China(No.6171177)National Key Scientific Instrument and Equipment Development Project(No.2013YQ240803)
文摘Ring artifact is the major factor that seriously influence quality of computed tomography(CT)image reconstruction,especially in testing large-scale workpieces.To remove ring artifact quickly and efficiently,a kind of ring artifact correction method is improved based on the post-processing CT image reconstruction in this paper.At first,transform the CT image from the rectangular coordinates into polar coordinates.Next,design multidimensional filter to filter the image and calculate the mean and variance of each filtered pixel in polar coordinates.The location of artifact point is determined by the double precision.One is the comparison of calculated variance and variance threshold,and another is the comparison of pixel value and pixel value threshold.Then,process the artifact points in a reasonable manner and do details remain to CT image in particular.At last,convert corrected polar image into rectangular coordinates.The actual experiment shows that compared with the original method,improved method can better correct the ring artifacts and keep the image details for CT images.It is a kind of practical ring artifact correction methods for subsequent processing and quantitative analysis.
文摘乳腺肿瘤是一种常见的恶性肿瘤,其临床诊断不但费时费力,还容易出现误诊.本文旨在建立一个基于乳腺数据自动分割的乳腺肿瘤计算机辅助诊断模型,提高临床诊断的速度和准确率.为了用卷积神经网络U-Net模型分割对比增强锥光束乳腺计算机断层扫描(Contrast-Enhanced Cone-Beam Breast CT,CE-CBBCT)数据,本文首先沿冠状面将3维数据转换成2维切片,通过数据默认的窗口对其进行归一化处理.实验结果显示,使用U-Net卷积神经网络对数据进行分割,Dice系数和IoU(Intersection over Union)分别为0.7920和0.6962.然后,本文用不同骨干网络(即各种深度学习分类网络)去替换U-Net的编码器,再次进行分割并对比不同特征提取对分割性能的影响,发现旋转增广方式可以提升各分割网络的性能.其中,基于ResNet152的U形分割网络的性能最好,Dice系数和IoU分别达到0.8410和0.7576.随后,本文又在所有模型中选取5个性能最好的模型组成一个集成模型,重复分割实验,发现此模型有最佳分割性能,平均Dice系数和IoU可达0.8463和0.7676,性能显著提升.值得指出的是,在处理数据时本文仅使用数据默认的窗口,降低了对人工的依赖.
文摘Aim To investigate the relationship between the positioning of the lower central incisor and physical morphology of the surrounding alveolar bone. Methodology Thirty-eight patients （18 males, 20 females）, with mean age of 13.4 years, were included in this study. As part of orthodontic treatment planning the patients were required to take dental Cone-beam CT （CBCT） covering the region of lower incisors, the sur- rounding alveolar bone and the mandibular symphysis. The cephalometric parameters were designed and measured to indicate the inclination of lower central incisor and physical morphology of the adjacent alveolar bone. Computer-aided descriptive statistical analysis was performed using SPSS 15.0 software package for Windows. A correlation analysis and a linear regression analysis between the incisor inclination and the alveolar bone morphology were performed. Results Significant positive correlations were found between the lower central incisor inclination and the morphological contour of the alveolar bone （P〈0.05）. The lower central incisor root apex was closer to the lingual alveolar crest when it was buccally inclined. Conclusion The morphology of the alveolar bone may be affected by incisal inclination.
文摘基于实验室设备自主开发了一套320 kV锥束工业CT成像系统。首先设计了基于运动控制卡的六轴机械扫描平台,并将其与320 kV射线机、平板探测器结合搭建了锥束CT成像系统的硬件。其次,利用平板探测器和运动控制卡的动态链接库开发锥束CT投影数据采集软件和重建软件,自主开发FDK算法并基于CUDA架构实现FDK算法的GPU加速。最后,采用阻尼器对锥束CT成像系统进行测试。结果表明,整个系统运行稳定,可以获得高质量的CT重建图像,且重建效果与VG STUDIO MAX 3.4软件的重建效果一致。
基金supported by the National Natural Science Foundation of China(Nos.61771279,11435007)the National Key Research and Development Program of China(No.2016YFF0101304)
文摘Because of the growing concern over the radiation dose delivered to patients, X-ray cone-beam CT(CBCT) imaging of low dose is of great interest. It is difficult for traditional reconstruction methods such as Feldkamp to reduce noise and keep resolution at low doses. A typical method to solve this problem is using optimizationbased methods with careful modeling of physics and additional constraints. However, it is computationally expensive and very time-consuming to reach an optimal solution. Recently, some pioneering work applying deep neural networks had some success in characterizing and removing artifacts from a low-dose data set. In this study,we incorporate imaging physics for a cone-beam CT into a residual convolutional neural network and propose a new end-to-end deep learning-based method for slice-wise reconstruction. By transferring 3D projection to a 2D problem with a noise reduction property, we can not only obtain reconstructions of high image quality, but also lower the computational complexity. The proposed network is composed of three serially connected sub-networks: a cone-to-fan transformation sub-network, a 2D analytical inversion sub-network, and an image refinement sub-network. This provides a comprehensive solution for end-to-end reconstruction for CBCT. The advantages of our method are that the network can simplify a 3D reconstruction problem to a 2D slice-wise reconstruction problem and can complete reconstruction in an end-to-end manner with the system matrix integrated into the network design. Furthermore, reconstruction can be less computationally expensive and easily parallelizable compared with iterative reconstruction methods.
文摘The accuracy of conventional superposition or convolution methods for scatter correction in kV-CBCT is usually compromised by the spatial variation of pencil-beam scatter kernel (PBSK) due to finite size, irregular external contour and heterogeneity of the imaged object. This study aims to propose an analytical method to quantify the Compton single scatter (CSS) component of the PBSK, which dominates the spatial distribution of total scatter assuming that multiple scatter can be estimated as a constant background and Rayleigh scatter is the secondary source of scatter. The CSS component of PBSK is the line integration of scatter production by incident primary photons along the beam line followed by the post-scattering attenuation as the scattered photons traverse the object. We propose to separate the object-specific attenuation term from the line integration and equivalently replace it with an average value such that the line integration of scatter production is object independent but only beam specific. We derived a quartic function formula as an approximate solution to the spatial distribution of the unattenuated CSS component of PBSK. The “effective scattering center” is introduced to calculate the average attenuation. The proposed analytical framework to calculate the CSS was evaluated using parameter settings of the On-Board Imager kV-CBCT system and was found to be in high agreement with the reference results. The proposed method shows highly increased computational efficiency compared to conventional analytical calculation methods based on point scattering model. It is also potentially useful for correcting the spatial variant PBSK in adaptive superposition method.
