Supervised Domain Adaptation for Automated Semantic Segmentation of the Atrial Cavity

Atrial fibrillation (AF) is the most common cardiac arrhythmia. At present, cardiac ablation is the main treatment procedure for AF. To guide and plan this procedure, it is essential for clinicians to obtain patient-specific 3D geometrical models of the atria. For this, there is an interest in automatic image segmentation algorithms, such as deep learning (DL) methods, as opposed to manual segmentation, an error-prone and time-consuming method. However, to optimize DL algorithms, many annotated examples are required, increasing acquisition costs. The aim of this work is to develop automatic and high-performance computational models for left and right atrium (LA and RA) segmentation from a few labelled MRI volumetric images with a 3D Dual U-Net algorithm. For this, a supervised domain adaptation (SDA) method is introduced to infer knowledge from late gadolinium enhanced (LGE) MRI volumetric training samples (80 LA annotated samples) to a network trained with balanced steady-state free precession (bSSFP) MR images of limited number of annotations (19 RA and LA annotated samples). The resulting knowledge-transferred model SDA outperformed the same network trained from scratch in both RA (Dice equals 0.9160) and LA (Dice equals 0.8813) segmentation tasks.     

一种基于2D-CNN的激光超声表面缺陷检测方法

激光超声表面缺陷检测的过程中,缺陷的定量表征通常依赖于操作者的判断,易受到人为因素干扰,致使检测结果不稳定.针对这一问题,提出一种基于图像识别的二维卷积神经网络(2D-CNN)的缺陷自动分类检测方法.利用有限元方法模拟激光超声检测过程,并采集超声信号数据用于训练分类模型;使用连续小变换(CWT)处理超声信号得到小波时频...     

基于锥束CT序列图像的三维缺陷检测方法

根据锥束CT序列图像各向同性及缺陷实体在序列图像层间位移和变化较小的特点,提出了一种适合锥束CT序列图像的三维缺陷检测方法。首先结合多目标跟踪思想,利用三维连通区域标记算法提取三维缺陷实体,并建立缺陷对应关系哈希表,以解决缺陷检测中目标轨迹的分叉;然后根据噪声目标的固有特性,对虚假缺陷信息进行有效删除,最终可获得真实缺陷目标实体,缺陷提取精度可达到像素。通过对空心涡轮叶片蜡模锥束CT图像进行实验,结果表明。该方法能准确地提取有较强噪声影响的序列图像的三维缺陷。     

Polyspectral technique for the analysis of stress-waves characteristics and species recognition in wood veneers

In this work a simple technique to obtain information about the species of wood samples using stress-wave sounds in the audible range is presented. However, spectra of wood sounds generated by pendulum impact are very complex and feature extraction for classification purposes is very difficult. Polyspectral techniques have been successfully applied to several problems from radar pattern recognition to medical signal processing. Following this approach, convolution of four different sound impacts has been done. This permits to extract clear polyspectral features suitable for wood species recognition with possible applications to both human assisted and automatic wood identification systems with minimal intersample variability. Results indicate that using this technique only the two most intense polyspectral peaks are enough for species recognition.     

Enhancing the discrimination accuracy between metastases, gliomas and meningiomas on brain MRI by volumetric textural features and ensemble pattern recognition methods

Three-dimensional (3D) texture analysis of volumetric brain magnetic resonance (MR) images has been identified as an important indicator for discriminating among different brain pathologies. The purpose of this study was to evaluate the efficiency of 3D textural features using a pattern recognition system in the task of discriminating benign, malignant and metastatic brain tissues on T1 postcontrast MR imaging (MRI) series. The dataset consisted of 67 brain MRI series obtained from patients with verified and untreated intracranial tumors. The pattern recognition system was designed as an ensemble classification scheme employing a support vector machine classifier, specially modified in order to integrate the least squares features transformation logic in its kernel function. The latter, in conjunction with using 3D textural features, enabled boosting up the performance of the system in discriminating metastatic, malignant and benign brain tumors with 77.14%, 89.19% and 93.33% accuracy, respectively. The method was evaluated using an external cross-validation process; thus, results might be considered indicative of the generalization performance of the system to "unseen" cases. The proposed system might be used as an assisting tool for brain tumor characterization on volumetric MRI series.     

基于贝叶斯神经网络的近红外光谱实木地板表面缺陷检测

实木地板的表面缺陷直接影响其力学性能和产品等级,表面缺陷的快速检测对实木地板的在线分选具有重要的现实意义。针对视觉方法检测实木地板表面缺陷识别率低的问题,提出了一种基于近红外光谱分析技术的检测方法。首先,分别采集规格为200 mm×100 mm×20 mm的表面带有活节、死节以及无缺陷的实木地板的光谱数据各60份,其中30份作为训练样本,30份作为测试样本;其次,使用高斯滤波(GSF)、分段多元散射校正(PMSC)和去趋势法(DT)等方法对采集到的光谱数据进行预处理,降低光谱噪声、消除光谱的散射影响;然后,利用改进遗传算法从处理后的光谱中提取特征波长用于构建缺陷识别与分类模型;最后,使用基于贝叶斯理论改进的神经网络构建实木地板缺陷识别和分类模型。实验使用含有活节、死节以及无缺陷的实木地板样本对模型进行训练和测试,结果表明：通过贝叶斯神经网络构建的缺陷识别与分类模型能够准确识别活节、死节和无缺陷三类实木地板,识别率分别为92.20%, 94.47%和95.57%。证明了实木地板表面缺陷类型与其近红外吸收光谱密切相关,并为下一步实现实木地板表面缺陷的准确定位提供快速检测方法。     

Strategy for automatic and complete three-dimensional optical digitization

This Letter proposes a new strategy of a three-dimensional (3D) scanning pipeline to achieve complete 3D digitization of complex objects in a real scene. This strategy consists of a one-dimensional array of optical 3D sensors combined with an automatically controlled turntable. An efficient calibration method for the sensor array is presented to guarantee the accuracy of the 3D measurement. Furthermore, an automatic registration technique is also proposed for aligning multiple range images taken from sensor array. Experiment results are also presented to demonstrate the proposed approach.     

Wood species identification using improved active shape model

In this paper, we propose a robust wood species identification scheme by using color wood surface images. First, a novel wood image acquirement system is devised, and the wood color image is converted into a V₁V₂I color-base image. Second, the corresponding grey histograms for V₁ and V₂ are established. Third, an improved active shape model is used to fulfill the curve deformation of the histogram curve of the standard specimen. This active shape model will then converge to the histogram curve of the test specimen. Finally, wood recognition is performed by comparing the initial and final active shape models with the histogram curve of the test specimen. We have experimentally proved that this scheme improves the mean recognition accuracy to approximately 90% for 5 wood species and that it can also be applied to the Gaussian noisy images. Moreover, the recognition accuracy can be further improved by combining this scheme with the texture feature recognition.     

Glasses-free 3D viewing systems for medical imaging

In this work we show two different glasses-free 3D viewing systems for medical imaging: a stereoscopic system that employs a vertically dispersive holographic screen (VDHS) and a multi-autostereoscopic system, both used to produce 3D MRI/CT images. We describe how to obtain a VDHS in holographic plates optimized for this application, with field of view of 7 cm to each eye and focal length of 25 cm, showing images done with the system. We also describe a multi-autostereoscopic system, presenting how it can generate 3D medical imaging from viewpoints of a MRI or CT image, showing results of a 3D angioresonance image.     

Three-dimensional recognition of occluded objects by using computational integral imaging

We have proposed a method to recognize partially occluded three-dimensional (3D) objects by using 3D volumetric reconstruction integral imaging (II). An II system captures multiple perspectives of occluded objects by using a microlens array. The reconstruction of the occluded 3D scene and target recognition are done digitally to reduce the effects of the occlusion. To verify system performance, we have implemented an optimum filter for object recognition. Both two-dimensional (2D) images and 3D II volumetric reconstructed images are considered. The correlation results of occluded 3D images for volumetric reconstruction show substantial improvements compared with those for conventional 2D imaging of occluded images.     

基于ICT切片图像相邻层信息的缺陷自动识别方法

 现有的工业计算机断层成像（ICT）图像缺陷识方法中,多采用对单张图像进行孤立评判方法,此类方法未能考虑到单张图像在相邻层图像信息关联性,因而易将孤立的噪音视为缺陷,造成误判。为解决这一问题,提出一种基于序列ICT切片图像自动识别方法,该方法将识别过程分为两步：单张图像的潜在缺陷提取和相邻层图像缺陷的匹配。第一步运用传统方法识别出每张图像中所有潜在缺陷;第二步根据真缺陷在相邻层具有匹配关系而伪缺陷则相对孤立的特点,通过分步匹配的方法确定每张图像上所有潜在缺陷在相邻层图像上的匹配关系,区分出真伪缺陷。最后通过实例验证表明：利用该方法可以有效得提高真缺陷得识别率,降低误判率。     

高频信息矢量匹配实现异源图像配准

介绍了一种通过提取红外与可见图像高频信息形成矢量模型的方法,该模型可用于可见和红外两种不同波段相同场景图像的配准。分析了红外与可见图像的共同特性,即边缘高频特性,并在作为模板的图像中提取这类高频信息。利用人工干预的方法形成矢量模型,该矢量模型可通过实时姿态信息进行实时三维变换,并重新投影为二维图形,据此可在另一图像中进行特征搜索,达到两者匹配的目的。利用实际数据进行了实验分析,结果显示,利用本文算法提取的矢量模型,经变换后能可靠地实现异源目标自动识别;实验证明此方法有效可行,并可推广应用于各种异源图像间的自动目标识别和配准处理。     

Wavelet-based characterization of vertebral trabecular bone structure from magnetic resonance images at 3 T compared with micro-computed tomographic measurements

Trabecular bone structure and bone density contribute to the strength of bone and are important in the study of osteoporosis. Wavelets are a powerful tool in characterizing and quantifying texture in an image. The purpose of this study was to validate wavelets as a tool in computing trabecular bone thickness directly from gray-level images. To this end, eight cylindrical cores of vertebral trabecular bone were imaged using 3-T magnetic resonance imaging (MRI) and micro-computed tomography (microCT). Thickness measurements of the trabecular bone from the wavelet-based analysis were compared with standard 2D structural parameters analogous to bone histomorphometry (MR images) and direct 3D distance transformation methods (microCT images). Additionally, bone volume fraction was determined using each method. The average difference in trabecular thickness between the wavelet and standard methods was less than the size of 1 pixel size for both MRI and microCT analysis. A correlation (R) of .94 for microCT measurements and that of .52 for MRI were found for the bone volume fraction. Based on these results, we conclude that wavelet-based methods deliver results comparable with those from established MR histomorphometric measurements. Because the wavelet transform is more robust with respect to image noise and operates directly on gray-level images, it could be a powerful tool for computing structural bone parameters from MR images acquired using high resolution and thus limited signal scenarios.     

Metastable ringlike semitopological solitons

We show the existence of new stable ringlike localized scalar field configurations whose stability is due to a combination of topological and nontopological charges. In that sense these defects may be called semitopological. These rings are Noether charged and also carry Noether current (they are superconducting). They are local minima of the energy in scalar field theories with an unbroken U(1) global symmetry. We obtain numerical solutions of the field configuration corresponding to large rings and derive virial theorems demonstrating their stability. We also derive the minimum energy field configurations in 3D and simulate the evolution of a finite size Q ring on a three dimensional lattice.     

Ultrasonic techniques for nondestructive testing of standing trees

The aim of this study is to assess wood quality of standing trees with ultrasonic velocity methods. The ultrasonic techniques currently used are: scattering based techniques that use travel time and other wave parameters for measuring the elastic constants of wood, and ultrasonic tomographic imaging techniques which seek to provide a high resolution picture of the defect. Scattering based techniques were firstly used as nondestructive techniques with increment cores of 5 mm diameter. On these specimens measurements with bulk waves--longitudinal and shear were used to detect the influence of sylvicultural treatment (pruning) on wood quality and also to detect defects (slope of grain, wavy figures). High resolution imaging techniques were developed for imaging abnormalities induced by biological attacks of fungi or insects in standing trees. In this case, ultrasonic tomography refers to cross-sectional imaging of trees from data collected by illuminating the tree from different directions. Ultrasonic images were reconstructed from the time of flight. The images were reconstructed from 120 measurements in situ and through the bark with direct transmission technique, using 1 MHz transducers. The resolution is 5 cm.     

Time-of-flight remote detection MRI of thermally modified wood

We demonstrate that time-of-flight (TOF) remote detection (RD) magnetic resonance imaging (MRI) provides detailed information about physical changes in wood due to thermal modification that is not available with conventional nuclear magnetic resonance (NMR) based techniques. In the experiments, xenon gas was forced to flow through Pinus sylvestris pine wood samples, and the flow paths and dispersion of gas atoms were observed by measuring ¹²⁹Xe TOF RD MRI images from the samples. MRI sensitivity of xenon was boosted by the spin exchange optical pumping (SEOP) method. Two different samples were studied: a reference sample, dried at low temperature, and a modified sample, which was thermally modified at 240 °C after the drying. The samples were taken next to each other from the same wood plank in order to ensure the comparability of the results. The most important conclusion is that both the smaller dispersion observed in all the TOF RD experiments independent of each other and the decreased amount of flow paths shown by the time projection of z-encoded TOF RD MRI experiment imply that a large amount of pits connecting tracheid cells are closed in thermal modification. Closed pits may be one reason for reduced moisture content and improved dimensional stability of wood achieved in thermal modification. This is the first time biological samples have been investigated by TOF RD MRI.     

MR imaging of flow with locally high spatial resolution.

Locally focused magnetic resonance imaging (LF MRI) allows imaging with variable spatial resolution within the field of view (FOV). Because LF MRI uses a priori information to provide locally high resolution in regions with rapid spatial variations in intensity (e.g., blood/tissue interface), it allows accurate reproduction of intense sharp edges in the specimen without blurring and truncation artifacts. This study employs LF MRI for 3D imaging of stationary and pulsatile flow. In the implemented version of LF MRI analytically defined basis functions are used to determine image intensity in regions depicted with low or high resolution. It is demonstrated that LF MRI of flow allows a significant (i.e. 3-4 times) reduction in scan time as compared to conventional FT MRI. It is also shown that LF images of pulsatile flow have a decreased appearance of ghosting artifacts as compared to the images reconstructed by using the conventional method.     

Automatic registration of serial sections of mouse lymph node by using Image-Reg

The first step towards the three-dimensional (3D) reconstruction of histological structures from serial sectioned tissue blocks is the proper alignment of microscope image sequences. We have accomplished an automatic rigid registration program, named Image-Reg, to align serial sections from mouse lymph node and Peyer's patch. Our approach is based on the calculation of the pixel-correlation of objects in adjacent images. The registration process is mainly divided into two steps. Once the foreground images have been segmented from the original images, the first step (primary alignment) is performed on the binary images of segmented objects; this process includes rotation by using the moments and translation through the X, Y axes by using the centroid. In the second step, the matching error of two binary images is calculated and the registration results are refined through multi-scale iterations. In order to test the registration performance, Image-Reg has been applied to an image and its transformed (rotated) version and subsequently to an image sequence of three serial sections of mouse lymph node. In addition, to compare our algorithm with other registration methods, three other approaches, viz. manual registration with Reconstruct, semi-automatic landmark registration with Image-Pro Plus and the automatic phase-correlation method with Image-Pro Plus, have also been applied to these three sections. The performance of our program has been also tested on other two-image data sets. These include: (a) two light microscopic images acquired by the automatic microscope (stitched with other software); (b) two images fluorescent images acquired by confocal microscopy (tiled with other software). Our proposed approach provides a fast and accurate linear alignment of serial image sequences for the 3D reconstruction of tissues and organs.     

支持向量机复合核函数的高光谱显微成像木材树种分类

采用体视显微高光谱成像方法,构建木材树种分类识别模型。利用SOC710VP体视显微高光谱图像采集系统获取可见光/近红外（372.53～1 038.57 nm）波段内的木材高光谱图像。首先,采用ENVI软件提取木材样本感兴趣区域（ROI）的平均光谱,分别采用连续投影算法（SPA）和竞争性自适应重加权算法(CARS)对光谱数据进行降维。再利用支持向量机（SVM）分别建立木材样本采集波段和特征波长下的分类模型。然后,在空间维采用第一主成分图像,计算基于灰度共生矩阵（GLCM）的木材纹理特征。在0°,45°,90°和135°四个方向计算能量、熵、惯性矩、相关性等16个特征参数后输入SVM进行木材树种分类处理。最后,采用四个复合核函数SVM进行光谱维和空间维的特征融合及分类识别。20个树种的分类实验结果表明,CARS的特征波长选择效果和运行速度较好一些,采用普通SVM进行木材光谱维特征分类处理时,测试集分类准确率达到了92.166 7%。采用基于GLCM的木材空间维纹理特征时,采用普通SVM的测试集分类准确率是60.333 0%,具有较低的分类精度。在将光谱维和空间维纹理特征进行数据融合及分类处理时,采用复合核函数SVM分类具有更好的效果。采用第二个复合核函数的SVM分类精度最高,测试集分类正确率是94.166 7%,运行时间为0.254 7 s。另外,采用第一个和第三个复合核函数的SVM的测试集分类准确率分别是93.333 3%和92.610 0%,运行时间分别为0.180 0和0.260 2 s。可以看出,采用这3种复合核函数的SVM进行木材树种分类,分类精度都高于采用普通SVM的光谱维或者空间维的分类识别精度。因此,利用体视显微高光谱成像和复合核函数SVM可以提高木材树种分类精度,为木材树种快速分类提供了参考。     

视频图像记录判读系统

光电经纬仪的摄影胶片图像和CCD摄像机的视频图像的数字化处理是兵器试验靶场中始终未能很好解决的问题。主要源于两个问题:一是视频图像的数字化采集和转换;二是数字图像的自动处理。介绍了根据靶场广泛应用的光学图像记录、判读需求而研制的视频图像记录判读系统,并对所采用的视频图像实时存储技术、字符识别技术、目标识别与定位技术等几项主要的关键技术进行了说明,并给出了应用结果。结果表明,图像采集记录灵活方便,速度快,不丢帧;目标图像处理精度高,实时性强,自动处理能力好,使光测数据处理的精度和效率都有了很大的提高。