基于机器学习方法的前列腺癌DWI多参数分析及其应用

探究使用机器学习方法,提升对扩散加权成像(DWI)多参数图的前列腺癌(PCa)诊断的准确性。对39例前列腺癌患者、56例良性患者,进行磁共振扩散加权图像的采集,并使用传统单指数模型(Mono)、拉伸指数模型(SEM)、弥散张量成像(DTI)模型、弥散峰度成像(DKI)模型以及体内素不相干运动扩散(IVIM)模型等5种重建模型,得到共计16个参数图,而后对于每一个参数图进行直方图分析,得到相关图像特征后使用机器学习的方法进行分类。 使用支持向量机和随机森林两种分类器对前列腺病变进行良恶性分类,随机森林分类器的AUC值可以达到0.98,具有较高的分类性能。另外,对特征进行重要性排序后,发现DKI参数图是肿瘤分类的重要指标。     

基于深度学习单投影的CT断层成像三维重建

【摘要】目的:提出一种基于深度学习的计算机断层扫描（CT）单视图断层成像三维（3D）重建方法,在减少数据采集量和降低成像剂量的情况下对不同患者进行CT图像的3D重建。方法:对不同患者的CT图像进行数据增强和模拟生成对应的数字重建放射影像（DRR）,并进行数据归一化操作。利用预处理后的数据通过卷积神经网络训练出一个普适于不同患者的神经网络模型。将训练好的神经网络模型部署在测试数据集上,使用平均绝对误差（MAE）、均方根误差（RMSE）、结构相似性（SSIM）和峰值信噪比（PSNR）对重建结果进行评估。结果:定性和定量分析的结果表明,该方法可以使用不同患者的单张2D图像分别重建出质量较高的3D CT图像,MAE、RMSE、SSIM和PSNR分别为0.006、0.079、0.982、38.424 dB。此外,相比特定于单个患者的情况,该方法可以大幅度提高重建速度并节省70%的模型训练时间。结论:构建的神经网络模型可通过不同患者的2D单视图重建出相应患者的3D CT图像。因此,本研究对简化临床成像设备和放射治疗当中的图像引导具有重要作用。     

基于深度学习的低剂量CT图像去噪

目的:提出一种基于深度学习的方法用于低剂量CT（LDCT）图像的噪声去除。方法:首先进行滤波反投影重建,然后利用多尺度并行残差U-net（MPR U-net）的深度学习模型对重建后的LDCT图像进行去噪。实验数据采用LoDoPaB-CT挑战赛的医学CT数据集,其中训练集35 820张图像,验证集3 522张图像,测试集3 553张图像,并采用峰值信噪比（PSNR）与结构相似性系数（SSIM）来评估模型的去噪效果。结果:LDCT图像处理前后PSNR分别为28.80、38.22 dB,SSIM分别为0.786、0.966,平均处理时间为0.03 s。结论:MPR U-net深度学习模型能较好地去除LDCT图像噪声,提升PSNR,保留更多图像细节。     

磁共振部分K空间重建算法

磁共振成像（MRI）是必要的获取临床图像的影像学方法之一,但是它获取数据过程缓慢使得成像时间过长。目前提出了许多高效的成像算法来降低磁共振的成像时间,如半傅里叶成像和压缩感知MRI等。半傅里叶成像仅采用多于一半的K空间数据进行图像重建,不仅提高了MRI的成像速度,而且降低了运动伪影,是有效的部分K空间重建技术之一。基于压缩感知理论的MRI仅采用25%~30%的K空间数据就能重建出MRI图像,与其它成像技术相比,可在相同的扫描时间内获得更高质量的MRI图像,也可在相同的空间分辨率下加速成像。本文综述几种半傅里叶成像算法的原理,也阐述了压缩感知理论与MRI相结合的原理,包括MR图像的稀疏表示、K空间的采样轨迹设计、重建算法的选择等。     

PM滤波方法对提高不同b值弥散张量成像图像质量的应用研究

目的在不同b值条件下,研究PM滤波方法对提高弥散张量成像（DTI）技术图像质量的作用。方法在b值为800～2800s／mm2的条件下分别对模体及志愿者实施DTI．利用改进的PM滤波方法对获得的弥散加权成像（DWI）图像进行处理后求解张量参数．得到反映水分子各向异性扩散程度的FA图。通过比较标准FA图与处理后FA图间的均方根误差（RMSE）．评价PM滤波的效果。结果随着b值的升高,图像信噪比逐渐降低．FA误差逐渐增大。当b值小于2000s／mm2时．经过PM滤波后．可得到RMSE较低的后处理图像。结论b值在1000．2000s／mm2范围内,应用PM滤波方法是提高人脑DTI图像质量的一种有效手段．     

基于扩散张量的脑白质内神经纤维束的可视化技术

介绍了在功能磁共振成像的研究中发展非常快的扩散张量成像技术的基本原理，以及如何利用扩散张量数据来重建脑白质内的神经纤维柬图像。其中主要介绍了白质束成像技术及其优缺点，并且分析了神经纤维束可视化技术的应用前景及其局限性。     

应用DTI技术对人脑白质纤维束的初步研究

目的建立扩散张量纤维束成像对人脑白质纤维的显示方法,并应用中国数字化可视人体数据进行对照观察,验证扩散张量成像(DTI)方法的可靠性。方法选择5名健康志愿者进行DTI成像,采用DtiStudio软件进行分析处理,重建出部分各向异性(FA)图、容积比(VR)图、相对各向异性(RA)图、表面扩散系数(ADC)图以及二维彩色张量图。应用中国数字化可视人体数据集断面图像、FA图及彩色FA图进行对照观察,利用fibertracking纤维跟踪软件及3DMRI软件进行三维重建显示脑内主要白质纤维束,辨认脑内白质纤维束的位置、形态。结果应用DTI纤维束成像可以清晰准确地描绘脑白质内主要神经纤维束的解剖图谱,包括联络纤维如弓形纤维、钩束、扣带束、上纵束和下纵束,连合纤维如胼胝体、前连合和穹隆,投射纤维如锥体束、视放射、内侧丘系等。DTI纤维束成像结果与已知解剖知识、中国可视化人体断面图像具有很好的一致性。结论应用DTI纤维束成像可以清晰准确地描绘脑白质内主要神经纤维束的解剖图谱,其结果与中国可视化人体断面图像、已知解剖知识是一致的,应用DTI纤维束成像研究脑内纤维连通性是可靠的。     

基于小波稀疏的磁性纳米粒子成像算法研究

目的 为了实现磁性纳米粒子成像(magnetic particle imaging,MPI)中粒子浓度空间分布的快速精准成像,针对系统矩阵成像方法所构建矩阵方程的求解问题,本文提出一种基于小波稀疏的MPI算法.方法 首先通过仿真从基于零场线的开放式MPI电磁系统中获得MPI信号构建矩阵方程;然后在经典代数重建算法(algebraic reconstruction technique,ART)每次迭代后均采用小波变换提取图像中粒子分布边缘的非平稳特征,结合阈值算子稀疏运算去除图像中的干扰信号,实现粒子浓度空间分布成像;最后用峰值信噪比参数(peak signal-to-noise ratio,PSNR)对不同噪声下的成像结果进行分析.结果 当系统信噪比为30 dB、20 dB、10 dB时,基于小波稀疏的MPI算法在快速收敛的前提下,所成图像的PSNR参数相较经典代数重建算法分别提升了67.83％、18.66％、8.05％.结论 在低噪声水平下,基于小波稀疏的MPI算法可在短时间内实现粒子分布状况的高质量成像.     

基于压缩感知的磁共振图像重建算法的VC实现

我们对压缩感知重建算法在MRI中的应用进行了研究,并在VC6.0平台下对其进行工程实现。该程序主要由以下几部分组成：（1）建立图像模型及引入仪器采集数据;（2）设计采样方法模拟压缩感知的稀疏采集;（3）选择图像稀疏变换方法;（4）选用压缩感知算法重建得到图像。结果表明：使用该方法可以成功重建出高质量的图像,为压缩感知算法在MRI扫描仪器上的使用提供了重要参考。     

一种新的磁共振扩散成像实验参数控制方法

扩散敏感梯度磁度的方向及强度是磁共振扩散成像实验的重要参数，但这二个参数不能由用户通过设备自带的软件设定。本文介绍一种新的方法，通过修改MRI扫描机内部的数据文件，用户可以方便与精确地设定扩散加权成像DWI及扩散张景成像DTI的实验参数，而且可以为MRI扫描机增加新的功能。     

A model‐based reconstruction for undersampled radial spin‐echo DTI with variational penalties on the diffusion tensor

Radial spin‐echo diffusion imaging allows motion‐robust imaging of tissues with very low T₂ values like articular cartilage with high spatial resolution and signal‐to‐noise ratio (SNR). However, in vivo measurements are challenging, due to the significantly slower data acquisition speed of spin‐echo sequences and the less efficient k‐space coverage of radial sampling, which raises the demand for accelerated protocols by means of undersampling. This work introduces a new reconstruction approach for undersampled diffusion‐tensor imaging (DTI). A model‐based reconstruction implicitly exploits redundancies in the diffusion‐weighted images by reducing the number of unknowns in the optimization problem and compressed sensing is performed directly in the target quantitative domain by imposing a total variation (TV) constraint on the elements of the diffusion tensor. Experiments were performed for an anisotropic phantom and the knee and brain of healthy volunteers (three and two volunteers, respectively). Evaluation of the new approach was conducted by comparing the results with reconstructions performed with gridding, combined parallel imaging and compressed sensing and a recently proposed model‐based approach. The experiments demonstrated improvements in terms of reduction of noise and streaking artifacts in the quantitative parameter maps, as well as a reduction of angular dispersion of the primary eigenvector when using the proposed method, without introducing systematic errors into the maps. This may enable an essential reduction of the acquisition time in radial spin‐echo diffusion‐tensor imaging without degrading parameter quantification and/or SNR. Copyright © 2015 John Wiley & Sons, Ltd.     

融合白质fMRI的dMRI纤维追踪重建研究

基于弥散磁共振成像(dMRI)的纤维束重建,是分析大脑白质结构的主要工具.现有的纤维追踪成像算法受dMRI分辨率及成像机理约束,在构建大脑白质灰质边界区域的纤维时成像性能和准确性大大下降.为克服该缺陷,提出一种结合功能磁共振成像(fMRI)的新型dMRI纤维追踪成像算法.该算法引入表征白质中fMRI信号各向异性的空间相...     

Prior image constrained compressed sensing (PICCS): a method to accurately reconstruct dynamic CT images from highly undersampled projection data sets

When the number of projections does not satisfy the Shannon/Nyquist sampling requirement, streaking artifacts are inevitable in x-ray computed tomography (CT) images reconstructed using filtered backprojection algorithms. In this letter, the spatial-temporal correlations in dynamic CT imaging have been exploited to sparsify dynamic CT image sequences and the newly proposed compressed sensing (CS) reconstruction method is applied to reconstruct the target image sequences. A prior image reconstructed from the union of interleaved dynamical data sets is utilized to constrain the CS image reconstruction for the individual time frames. This method is referred to as prior image constrained compressed sensing (PICCS). In vivo experimental animal studies were conducted to validate the PICCS algorithm, and the results indicate that PICCS enables accurate reconstruction of dynamic CT images using about 20 view angles, which corresponds to an under-sampling factor of 32. This undersampling factor implies a potential radiation dose reduction by a factor of 32 in myocardial CT perfusion imaging.     

Diffusion-weighted imaging of the entire spinal cord

In spite of their diagnostic potential, the poor quality of available diffusion-weighted spinal cord images often restricts clinical application to cervical regions, and improved spatial resolution is highly desirable. To address these needs, a novel technique based on the combination of two recently presented reduced field-of-view approaches is proposed, enabling high-resolution acquisition over the entire spinal cord. Field-of-view reduction is achieved by the application of non-coplanar excitation and refocusing pulses combined with outer volume suppression for removal of unwanted transition zones. The non-coplanar excitation is performed such that a gap-less volume is acquired in a dedicated interleaved slice order within two repetition times. The resulting inner volume selectivity was evaluated in vitro. In vivo diffusion tensor imaging data on the cervical, thoracic and lumbar spinal cord were acquired in transverse orientation in each of four healthy subjects. An in-plane resolution of 0.7 x 0.7 mm(2) was achieved without notable aliasing, motion or susceptibility artifacts. The measured mean +/- SD fractional anisotropy was 0.69 +/- 0.11 in the thoracic spinal cord and 0.75 +/- 0.07 and 0.63 +/- 0.08 in cervical and lumbar white matter, respectively.     

Comparison of basis functions and q‐space sampling schemes for robust compressed sensing reconstruction accelerating diffusion spectrum imaging

Time constraints placed on magnetic resonance imaging often restrict the application of advanced diffusion MRI (dMRI) protocols in clinical practice and in high throughput research studies. Therefore, acquisition strategies for accelerated dMRI have been investigated to allow for the collection of versatile and high quality imaging data, even if stringent scan time limits are imposed. Diffusion spectrum imaging (DSI), an advanced acquisition strategy that allows for a high resolution of intra‐voxel microstructure, can be sufficiently accelerated by means of compressed sensing (CS) theory. CS theory describes a framework for the efficient collection of fewer samples of a data set than conventionally required followed by robust reconstruction to recover the full data set from sparse measurements. For an accurate recovery of DSI data, a suitable acquisition scheme for sparse q‐space sampling and the sensing and sparsifying bases for CS reconstruction need to be selected. In this work we explore three different types of q‐space undersampling schemes and two frameworks for CS reconstruction based on either Fourier or SHORE basis functions. After CS recovery, diffusion and microstructural parameters and orientational information are estimated from the reconstructed data by means of state‐of‐the‐art processing techniques for dMRI analysis. By means of simulation, diffusion phantom and in vivo DSI data, an isotropic distribution of q‐space samples was found to be optimal for sparse DSI. The CS reconstruction results indicate superior performance of Fourier‐based CS‐DSI compared to the SHORE‐based approach. Based on these findings we outline an experimental design for accelerated DSI and robust CS reconstruction of the sparse measurements that is suitable for the application within time‐limited studies.     

基于非局部均值平滑的体素内纤维结构估计

为了克服球面反卷积法对扩散加权图像噪声敏感的问题,提出一种基于非局部均值平滑的体素内纤维结构估计方法,该方法建立了一种面向磁共振扩散加权图像的非局部均值方法,对磁共振扩散加权图像进行平滑后再采用球面反卷积法对体素内纤维结构进行估计,从而提高球面反卷积法的抗噪性能。数值仿真数据和仿真实体数据实验结果表明,与直接采用球面反卷积方法和对数据进行非局部均值平滑后使用球面反卷积相比,采用本研究提出的方法得到的体素内纤维结构的平均角度误差更小,且较少存在边缘模糊现象。