Numerical simulations of motion-insensitive diffusion imaging based on the distant dipolar field effects

Diffusion weighting in MRI is commonly achieved with the pulsed-gradient spin-echo (PGSE) method. When combined with spin-warping image formation, this method often results in ghosts due to the sample's macroscopic motion. It has been shown experimentally (Kennedy and Zhong, MRM 2004;52:1–6) that these motion artifacts can be effectively eliminated by the distant dipolar field (DDF) method, which relies on the refocusing of spatially modulated transverse magnetization by the DDF within the sample itself. In this report, diffusion-weighted images (DWIs) using both DDF and PGSE methods in the presence of macroscopic sample motion were simulated. Numerical simulation results quantify the dependence of signals in DWI on several key motion parameters and demonstrate that the DDF DWIs are much less sensitive to macroscopic sample motion than the traditional PGSE DWIs. The results also show that the dipolar correlation distance (d_c) can alter contrast in DDF DWIs. The simulated results are in good agreement with the experimental results reported previously.     

A comparison study of multishot vs. single-shot DWI-EPI in the neonatal brain: reduced effects of ghosting compared to adults

In the neonatal brain, it is important to use a fast imaging technique to acquire all diffusion weighted images (DWI) for apparent diffusion coefficient (ADC) calculation. Taking into account the occurrence of typical echo planar imaging (EPI) artifacts, we have investigated whether single-shot (SSh) or multishot (MSh) DWI-EPI should be preferred. In 14 neonates, 17 adult patients and 5 adult volunteers, DWIs are obtained both with SSh and MSh EPI. The occurrence of artifacts and their influence on the ADC are explored and further quantified using simulations and phantom studies. Two radiologists scored overall image quality and diagnosability of all images. Single-shot and MSh DWI-EPI scored equally well in neonates with respect to overall image quality and diagnosability. In newborns, more motion artifacts in MSh can be noticed while N/2-ghost artifacts in SSh occur less frequently than in adults. Both N/2-ghost and motion artifacts result in significant ADC abnormalities. There is a serious risk that these artifacts will be mistaken for genuine diffusion abnormalities. N/2-ghost artifacts are hardly noticed in the neonatal brain, which might be due to smaller cerebrospinal fluid (CSF) velocity than in adults. Apparent diffusion coefficient values in MSh are unreliable if motion occurs. We conclude that for ADC calculations in neonates SSh DWI-EPI is more reliable than MSh.     

基于荧光透射谱和高光谱图像纹理的茶叶病害预测研究

为了实现对茶叶病害的准确预测,避免病害特征提取过程中对茶叶的二次破坏,利用荧光透射技术对茶叶赤叶病叶片的荧光透射光谱特性展开研究。实验采集了健康茶叶叶片样本45个、赤叶病初期叶片样本60个及赤叶病中期叶片样本60个,并按照2∶1的比例划分成训练集和预测集样本数,通过荧光透射手段利用高光谱仪器采集这些叶片的原始荧光透射光谱。通过对这3组叶片样本平均光谱强度曲线的分析,证实了利用荧光透射光谱信息对这3种病害类型叶片进行分类的可行性。然后使用多项式平滑（savitzky-golay, S-G）方法对原始光谱进行平滑和降噪处理。最后采用竞争性自适应重加权抽样法（competitive adaptive reweighted sampling, CARS）对预处理后的光谱数据进行特征波长的选取。经过50次加权采样后,最终选取出4个特征波长,分别为：463,512,586和613 nm。为了最大化提取样本的病害特征信息,强化分类器输入病害特征值的典型性,使用高光谱反射技术,采集4个特征波长下的高光谱图像,分别使用2种不同的纹理提取算法提取病害叶片图像的纹理信息进行对比分析。首先利用灰度共生矩阵（GLCM）提取4幅图像的纹理信息,分别计算4个方向的灰度共生矩阵(0°,45°,90°及135°),然后计算5个共生矩阵的均值和方差。为了提高鲁棒性,取4幅图像纹理信息的平均值作为该叶片的纹理特征值,最终得到10个特征值。利用LBP（local binary patterns）算法获取特征波长下高光谱图像的纹理信息,并使用Uniform模式对LBP模型进行降维,最终每幅图像得到944个维度的LBP特征值,同样取4幅图像的平均值作为该叶片的LBP纹理特征值。最后通过极限学习机（ELM）分别建立特征光谱联合灰度共生矩阵纹理信息及LBP算子纹理信息的预测模型,由于模型的输入特征值不在一个量纲,首先对输入特征值进行归一化处理,然后再定义模型的输出标签,即健康叶片的预测模型输出为1,赤叶病早期为2,中期为3。实验测得基于CARS-GLCM-ELM模型的预测准确率为81.82%,基于CARS-LBP-ELM模型的预测准确率为85.45%,说明利用荧光透射光谱联合LBP算子纹理信息预测效果更好。由于没有达到预期效果,利用Softplus函数对ELM的隐含层激活函数进行了优化,替换掉原来的Sigmod函数,优化后的模型预测分类正确率达到92.73%,基本达到了预期效果。该研究将病害叶片的荧光光谱信息和对应特征波长下高光谱图像的纹理信息进行了融合,研究结果可为茶叶病害的快速、准确预测提供一定的参考价值。     

基于高光谱图像的即食海参新鲜度无损检测

新鲜度是即食海参加工品质调控和贮藏品质监控的关键指标。针对感官评定和现有理化检测无法满足即食海参产品大批量、标准化、工业化生产问题,提出了一种基于高光谱图像的即食海参新鲜度快速无损检测方法,通过图像主成分分析和波段比运算相结合,优选特征波长和图像;依据海参腐败机理,建立图像纹理特征与即食海参新鲜度等级间的关联模型,实现即食海参新鲜度无损、快速评价。首先针对高光谱图像巨大的数据量展开降维研究。根据即食海参体壁光谱吸收特性,以具有明显化学吸收特征的波长(474和985 nm)为分界点,获得包括全检测波段(400～1 000 nm)在内的六个待处理波段,通过分段图像主成分分析实现待测波段的优选,利用权重系数和波段比图像运算,最终将686和985 nm波段比图像确定为特征图像。面向特征图像的感兴趣区域(ROI),构建灰度共生矩阵(gray-level co-occurrence matrix, GLCM)、灰度梯度共生矩阵(gray-gradient co-occurrence matrix, GGCM)、改进的局部二元模式纹理描述子(local binary pattern,LBP),分别提取纹理参数作为输入,以挥发性盐基氮(total volatile basic nitrogen, TVB-N)检测为标准,建立经粒子群优化的BP 神经网络(back propagation,BP)即食海参新鲜度判别模型,新鲜度等级判别准确率分别为90%,95%和80%。结果表明,即食海参高光谱图像灰度梯度共生矩阵的纹理特征用于新鲜度判别效果较好。为即食海参新鲜度快速无损检测方法研究和仪器开发提供了理论基础和数据支持。     

基于Wedgelet的分层图像压缩

提出了一种分层图像压缩框架:图像=边缘轮廓+纹理。对原图像进行了一种自适应的多尺度Wedgelet分析,抽取并编码了图像的边缘轮廓。基于Wedgelet分析了在残差图像中引入的伪迹所具有的局部振荡特性,采用自适应局部余弦变换分析了以纹理为主要内容的残差图像,在将变换系数重组成与小波系数类似的树形结构后,采用零树编码获得了嵌入式码流。实验结果表明,该算法的重建图像质量优于SPIHT算法,在较好地保留原图像边缘轮廓和有效地减少边缘附近振铃伪迹的同时,较清晰的保留了原图像的纹理特征。     

多局部残差连接注意网络的图像去模糊

针对现有的基于卷积神经网络的图像去模糊算法存在图像纹理细节恢复不清晰的问题,提出了一种基于多局部残差连接注意网络的图像去模糊算法。首先,采用一个卷积层进行浅层特征提取;其次,设计了一种新的基于残差连接和并行注意机制的多局部残差连接注意模块,用于消除图像模糊并提取上下文信息;再次,采用一个基于扩张卷积的成对连接模块进行细节恢复;最后,利用一个卷积层重建清晰图像。实验结果表明：在GoPro数据集上的PSNR (peak signal to noise ratio)和SSIM (structure similarity)分别为31.83 dB、0.927 5,在定性和定量两方面都表明所提方法能够有效地恢复模糊图像的纹理细节,网络性能优于对比方法。     

Magnetic resonance image reconstruction using trained geometric directions in 2D redundant wavelets domain and non-convex optimization

Reducing scanning time is significantly important for MRI. Compressed sensing has shown promising results by undersampling the k-space data to speed up imaging. Sparsity of an image plays an important role in compressed sensing MRI to reduce the image artifacts. Recently, the method of patch-based directional wavelets (PBDW) which trains geometric directions from undersampled data has been proposed. It has better performance in preserving image edges than conventional sparsifying transforms. However, obvious artifacts are presented in the smooth region when the data are highly undersampled. In addition, the original PBDW-based method does not hold obvious improvement for radial and fully 2D random sampling patterns. In this paper, the PBDW-based MRI reconstruction is improved from two aspects: 1) An efficient non-convex minimization algorithm is modified to enhance image quality; 2) PBDW are extended into shift-invariant discrete wavelet domain to enhance the ability of transform on sparsifying piecewise smooth image features. Numerical simulation results on vivo magnetic resonance images demonstrate that the proposed method outperforms the original PBDW in terms of removing artifacts and preserving edges.     

Motion artifacts reduction in brain MRI by means of a deep residual network with densely connected multi-resolution blocks (DRN-DCMB)

Objective: Magnetic resonance imaging (MRI) acquisition is inherently sensitive to motion, and motion artifact reduction is essential for improving image quality in MRI. Methods: We developed a deep residual network with densely connected multi-resolution blocks (DRN-DCMB) model to reduce the motion artifacts in T1 weighted (T1W) spin echo images acquired on different imaging planes before and after contrast injection. The DRN-DCMB network consisted of multiple multi-resolution blocks connected with dense connections in a feedforward manner. A single residual unit was used to connect the input and output of the entire network with one shortcut connection to predict a residual image (i.e. artifact image). The model was trained with five motion-free T1W image stacks (pre-contrast axial and sagittal, and post-contrast axial, coronal, and sagittal images) with simulated motion artifacts. Results: In other 86 testing image stacks with simulated artifacts, our DRN-DCMB model outperformed other state-of-the-art deep learning models with significantly higher structural similarity index (SSIM) and improvement in signal-to-noise ratio (ISNR). The DRN-DCMB model was also applied to 121 testing image stacks appeared with various degrees of real motion artifacts. The acquired images and processed images by the DRN-DCMB model were randomly mixed, and image quality was blindly evaluated by a neuroradiologist. The DRN-DCMB model significantly improved the overall image quality, reduced the severity of the motion artifacts, and improved the image sharpness, while kept the image contrast. Conclusion: Our DRN-DCMB model provided an effective method for reducing motion artifacts and improving the overall clinical image quality of brain MRI.     

一种八谱段滤光片成像系统设计

针对多光谱滤光片阵列成像采样率低,原始(Raw)数据稀疏所导致的重建图像模糊,高频信息丢失等问题,提出了一种新八谱段滤光片阵列分布方案,利用基于邻域梯度延伸方法对光谱Raw图像进行重建.首先基于二叉树生成法,在重复排列的4×4阵列中设计了一种等空间概率比的八谱段滤光片分布方案;然后针对传感器直接获取的稀疏Raw图像,计算各谱段采样点的梯度信息,在保持图像结构特征和纹理信息的基础上,利用邻域采样点的像素值和梯度值对未采样点进行重建,从而获得完整的光谱图像信息;最后,基于已重建的八谱段光谱图像,采用伪逆矩阵法重构各像素位置的31波段光谱值.结果表明,相对于主流图像重建方法,本文算法提高了重建八谱段光谱图像的峰值信噪比、复合峰值信噪比,降低了光谱均方差,更好地保留了图像的纹理和边缘,有效降低了多光谱滤光片阵列成像中的颜色伪影和图像模糊等现象.     

Super-resolution image reconstruction based on Tukey data fusion and bilateral-total-variation regularization

Images of high-resolution are desired and often required in most photoelectronic imaging applications, and corresponding image reconstruction algorithm has became the frontier topics. On the basis of stochastic theory, a novel super-resolution image reconstruction algorithm based on Tukey norm data fusion and bilateral total variation regularization is proposed in this paper. The Tukey norm is employed for fusing the data of low-resolution frames and removing outliers in the data, and then aiming at the sickness of super-resolution reconstruction, the bilateral total variation regularization as a priori knowledge about the solution is incorporated to remove the artifacts from the final answer and improve the convergence rate. Simulated and real experiment results show that the proposed algorithm can improve the image resolution greatly and it is immune to noise and errors in motion and blur estimation.     

基于LBP和多层次结构的异质手背静脉身份识别

针对多源异质的手背静脉异质图像的识别研究,提出了基于LBP和多层次结构的识别算法;首先对图像做适当的预处理,然后将LBP特征提取算法编码的手背静脉纹理特征图像作为多层次结构的输入,通过多层次结构的逐层由具体到抽象的特征提取,得到的特征具有更大的鲁棒性;最后该算法在多源异质的手背静脉图像库得到的识别率比传统的算法识别率高,达到96.57%;进一步表明该算法能够较好地解决由于多源异质问题对手背静脉识别所造成的识别率低的影响。     

Image fusion through local feature extraction by using multi-scale top-hat by reconstruction operators

The purpose of image fusion is to combine useful image features of different original images into the final fusion image, which will produce one useful result image for different applications. One of the main difficulties of image fusion is extracting useful image features of different original images. In some cases, useful image features are local image features of the whole image. To efficiently extract local image features and produce an efficient fusion result, an image fusion algorithm based on the extracted local image features by using multi-scale top-hat by reconstruction operators is proposed in this paper. Firstly, multi-scale local feature extraction using multi-scale top-hat by reconstruction operators is discussed. Then, based on the extracted multi-scale local features of different original images, the useful image features for image fusion are constructed. Finally, the constructed useful image features for image fusion are combined into the final fusion image. Experimental results on different types of images show that, the proposed algorithm performs well for image fusion.     

Dynamic cardiac MRI reconstruction using motion aligned locally low rank tensor (MALLRT)

Various sparse transform models have been explored for compressed sensing-based dynamic cardiac MRI reconstruction from vastly under-sampled k-space data. Recently emerged low rank tensor model using Tucker decomposition could be viewed as a special form of sparse model, where the core tensor, which is obtained using high-order singular value decomposition, is sparse in the sense that only a few elements have dominantly large magnitude. However, local details tend to be over-smoothed when the entire image is conventionally modeled as a global tensor. Moreover, low rankness is sensitive to motion as spatiotemporal correlation is corrupted by spatial misalignment between temporal frames. To overcome these limitations, this paper presents a novel motion aligned locally low rank tensor (MALLRT) model for dynamic MRI reconstruction. In MALLRT, low rank constraint is enforced on image patch-based local tensors, which correspond to overlapping blocks extracted from the reconstructed high-dimensional image after group-wise inter-frame motion registration. For solving the proposed model, this paper presents an efficient optimization algorithm by using variable splitting and alternating direction method of multipliers (ADMM). MALLRT demonstrated promising performance as validated on one cardiac perfusion MRI dataset and two cardiac cine MRI datasets using retrospective under-sampling with various acceleration factors, as well as one prospectively under-sampled cardiac perfusion MRI dataset. Compared to four state-of-the-art methods, MALLRT achieved substantially better image reconstruction quality in terms of both signal to error ratio (SER) and structural similarity index (SSIM) metrics, and visual perception in preserving spatial details and capturing temporal variations.     

Analyzing diffusion tensor images with ghosting artifacts: the effects of direct and indirect normalization

The current study aims to assess the applicability of direct or indirect normalization for the analysis of fractional anisotropy (FA) maps in the context of diffusion-weighted images (DWIs) contaminated by ghosting artifacts. We found that FA maps acquired by direct normalization showed generally higher anisotropy than indirect normalization, and the disparities were aggravated by the presence of ghosting artifacts in DWIs. The voxel-wise statistical comparisons demonstrated that indirect normalization reduced the influence of artifacts and enhanced the sensitivity of detecting anisotropy differences between groups. This suggested that images contaminated with ghosting artifacts can be sensibly analyzed using indirect normalization.     

Wavelet-based edge correlation incorporated iterative reconstruction for undersampled MRI

Undersampling k-space is an effective way to decrease acquisition time for MRI. However, aliasing artifacts introduced by undersampling may blur the edges of magnetic resonance images, which often contain important information for clinical diagnosis. Moreover, k-space data is often contaminated by the noise signals of unknown intensity. To better preserve the edge features while suppressing the aliasing artifacts and noises, we present a new wavelet-based algorithm for undersampled MRI reconstruction. The algorithm solves the image reconstruction as a standard optimization problem including a ?₂ data fidelity term and ?₁ sparsity regularization term. Rather than manually setting the regularization parameter for the ?₁ term, which is directly related to the threshold, an automatic estimated threshold adaptive to noise intensity is introduced in our proposed algorithm. In addition, a prior matrix based on edge correlation in wavelet domain is incorporated into the regularization term. Compared with nonlinear conjugate gradient descent algorithm, iterative shrinkage/thresholding algorithm, fast iterative soft-thresholding algorithm and the iterative thresholding algorithm using exponentially decreasing threshold, the proposed algorithm yields reconstructions with better edge recovery and noise suppression.     

Correction of out-of-FOV motion artifacts using convolutional neural network

PurposeSubject motion during MRI scan can result in severe degradation of image quality. Existing motion correction algorithms rely on the assumption that no information is missing during motions. However, this assumption does not hold when out-of-FOV motion happens. Currently available algorithms are not able to correct for image artifacts introduced by out-of-FOV motion. The purpose of this study is to demonstrate the feasibility of incorporating convolutional neural network (CNN) derived prior image into solving the out-of-FOV motion problem.Methods and materialsA modified U-net network was proposed to correct out-of-FOV motion artifacts by incorporating motion parameters into the loss function. A motion model based data fidelity term was applied in combination with the CNN prediction to further improve the motion correction performance. We trained the CNN on 1113 MPRAGE images with simulated oscillating and sudden motion trajectories, and compared our algorithm to a gradient-based autofocusing (AF) algorithm in both 2D and 3D images. Additional experiment was performed to demonstrate the feasibility of transferring the networks to different dataset. We also evaluated the robustness of this algorithm by adding Gaussian noise to the motion parameters. The motion correction performance was evaluated using mean square error (NMSE), peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM).ResultsThe proposed algorithm outperformed AF-based algorithm for both 2D (NMSE: 0.0066 ± 0.0009 vs 0.0141 ± 0.008, P < .01; PSNR: 29.60 ± 0.74 vs 21.71 ± 0.27, P < .01; SSIM: 0.89 ± 0.014 vs 0.73 ± 0.004, P < .01) and 3D imaging (NMSE: 0.0067 ± 0.0008 vs 0.070 ± 0.021, P < .01; PSNR: 32.40 ± 1.63 vs 22.32 ± 2.378, P < .01; SSIM: 0.89 ± 0.01 vs 0.62 ± 0.03, P < .01). Robust reconstruction was achieved with 20% data missed due to the out-of-FOV motion.ConclusionIn conclusion, the proposed CNN-based motion correction algorithm can significantly reduce out-of-FOV motion artifacts and achieve better image quality compared to AF-based algorithm.     

Facial expression recognition algorithm using LGC based on horizontal and diagonal prior principle

The traditional Local Binary Pattern (LBP) algorithm can analyze the center pixel and neighboring pixels of the gray relationship, using in facial expression recognition, but you cannot consider the eyes, mouth, forehead and other areas in the expression feature different trends in the gradient direction. Firstly, we propose the Local Gradient Coding (LGC) algorithm, though the binary encoding to the horizontal, vertical and diagonal gradients respectively, to produce the fusion characteristic, then this can fully describe the facial muscles texture, wrinkles and other local deformation of contains the expression information. On the other hand, in order to reduce the computational complexity, and to remove the redundant, while not lose the main information contained in the face texture expression. This paper proposes and optimizes a new LGC operator based on horizontal and diagonal gradient prior principle (LGC-HD). The experimental results from JAFFE database show that, LGC-HD algorithm is more quickly and effectively to extract facial expression feature than LGC algorithm. Comparing to the traditional LBP algorithm, LBP uniform pattern and Gabor filtering, this LGC-HD algorithm has a significant advantage in the recognition accuracy and run time.     

An improved Richardson-Lucy algorithm based on local prior

Ringing is one of the most common disturbing artifacts in image deconvolution. With a totally known kernel, the standard Richardson-Lucy (RL) algorithm succeeds in many motion deblurring processes, but the resulting images still contain visible ringing. When the estimated kernel is different from the real one, the result of the standard RL iterative algorithm will be worse. To suppress the ringing artifacts caused by failures in the blur kernel estimation, this paper improves the RL algorithm based on the local prior. Firstly, the standard deviation of pixels in the local window is computed to find the smooth region and the image gradient in the region is constrained to make its distribution consistent with the deblurring image gradient. Secondly, in order to suppress the ringing near the edge of a rigid body in the image, a new mask was obtained by computing the sharp edge of the image produced using the first step. If the kernel is large-scale, where the foreground is rigid and the background is smoothing, this step could produce a significant inhibitory effect on ringing artifacts. Thirdly, the boundary constraint is strengthened if the boundary is relatively smooth. As a result of the steps above, high-quality deblurred images can be obtained even when the estimated kernels are not perfectly accurate. On the basis of blurred images and the related kernel information taken by the additional hardware, our approach proved to be effective.     

Robust visual tracking base on adaptively multi-feature fusion and particle filter

In order to avoid the tracking failure based on single feature under the conditions of cluttered backgrounds illumination changes, a robust tracking algorithm was proposed based on adaptively multi-feature fusion and particle filter. Color histogram was used to describe the overall distribution characteristics of the target and histogram of oriented gradients containing some construction information and LBP is very effective to describe the image texture features. The Three features were fused in the frame of particle filter. Meanwhile, the weights of each feature were adjusted dynamically. The experimental results show that with adaptive fusion, the tracker becomes more robust to illumination changes, pose variations, partial occlusions, cluttered backgrounds and camera motion.     

Steganalysis for GIF images based on colors-gradient co-occurrence matrix

A steganalysis algorithm based on colors-gradient co-occurrence matrix (CGCM) is proposed in this paper. CGCM is constructed with colors matrix and gradient matrix of the GIF image, and 27-dimensional statistical features of CGCM, which are sensitive to the color-correlation between adjacent pixels and the breaking of image texture, are extracted. Support vector machine (SVM) technique takes the 27-dimensional statistical features to detect hidden message in GIF images. Experimental results indicate that the proposed algorithm is more effective than Zhao's algorithm for several existing GIF steganographic algorithms and steganography tools, especially for multibit assignment (MBA) steganography and EzStego. Furthermore, the time efficiency of the proposed algorithm is much higher than Zhao's algorithm.