Automated atlas-based segmentation of NISSL-stained mouse brain sections using supervised learning

The problem of segmentation of mouse brain images into anatomical structures is an important stage of practically every analytical procedure for these images. The present study suggests a new approach to automated segmentation of anatomical structures in the images of NISSL-stained histological sections of mouse brain. The segmentation algorithm is based on the method of supervised learning using the existing anatomical labeling of the corresponding sections from a specialized mouse brain atlas. A mouse brain section to be segmented into anatomical structures is preliminarily associated with a section from the mouse brain atlas displaying the maximum similarity. The image of this section is then preprocessed in order to enhance its quality and to make it as close to the corresponding atlas image as possible. An efficient algorithm of luminance equalization, an extension of the well-known Retinex algorithm is proposed. A random forest is trained on pixel feature vectors constructed based on the atlas section images and the corresponding class labels associated with anatomical structures extracted from the atlas anatomical labeling. The trained classifier is then applied to classify pixels of an experimental section into anatomical structures. A new combination of features based on superpixels and location priors is suggested. Accuracy of the obtained result is increased by using Markov random field. Procedures of luminance equalization and subsequent segmentation into anatomical structures have been tested on real experimental sections.     

Multiscale Segmentation of Three-Dimensional MR Brain Images

Segmentation of MR brain images using intensity values is severely limited owing to field inhomogeneities, susceptibility artifacts and partial volume effects. Edge based segmentation methods suffer from spurious edges and gaps in boundaries. A multiscale method to MRI brain segmentation is presented which uses both edge and intensity information. First a multiscale representation of an image is created, which can be made edge dependent to favor intra-tissue diffusion over inter-tissue diffusion. Subsequently a multiscale linking model (the hyperstack) is used to group voxels into a number of objects based on intensity. It is shown that both an improvement in accuracy and a reduction in image post-processing can be achieved if edge dependent diffusion is used instead of linear diffusion. The combination of edge dependent diffusion and intensity based linking facilitates segmentation of grey matter, white matter and cerebrospinal fluid with minimal user interaction. To segment the total brain (white matter plus grey matter) morphological operations are applied to remove small bridges between the brain and cranium. If the total brain is segmented, grey matter, white matter and cerebrospinal fluid can be segmented by joining a small number of segments. Using a supervised segmentation technique and MRI simulations of a brain phantom for validation it is shown that the errors are in the order of or smaller than reported in literature.     

Full anatomical labeling of magnetic resonance images of human brain by registration with multiple atlases

The problem of automatic segmentation of magnetic resonance (MR) images of human brain into anatomical structures is considered. Currently, the most popular segmentation algorithms are based on the registration (matching) of the input image with (to) an atlas—an image for which an expert labeling is known. Segmentation on the basis of registration with multiple atlases allows one to better take into account anatomical variability and thereby to compensate, to some extent, for the errors of matching to each individual atlas. In this work, a more efficient (in speed and memory) implementation is proposed of one of the best multiatlas label fusion algorithms in order to obtain a labeling of the input image. The algorithm is applied to the problem of segmentation of brain MR images into 43 anatomical regions with the use of the publicly available IBSR database, in contrast to the original work, where the authors provide test results for the problem of extraction of a single anatomical structure, the hippocampus.     

Multiscaled combination of MR and SPECT images in neuroimaging: A simplex method based variable-weight fusion

Single-photon emission computed tomography (SPECT) images alone are difficult to understand in diagnosis, since anatomical structures are absent from the data. Studies on combination attempt to locate functional changes of the SPECT image by the magnetic resonance (MR) image. Due to the low similarity between original images, fused results are always darkened, obscured or loss some crucial anatomical structures. This paper has solved these problems by the variable-weight matrix which is estimated by minimizing the cost function using the simplex method. Under the generalized intensity-hue-saturation (GIHS) framework, the multiscaled analysis is presented for a better detail preservation. Besides, interactive approaches are discussed for the gradual variation between original images and the control of detail performance. The similarity assessment evaluates several different methods on a normal brain atlas. Two clips show the interactive property of the proposed method, while two medical cases demonstrate its clinical values. We conclude that the proposed method is superior to traditional methods, when considering the definition and the information capacity of fused results.     

Reversible Interpolation of Vectorial Images by an Anisotropic Diffusion-Projection PDE

In this paper, a nonlinear model for the interpolation of vector-valued images is proposed. This model is based on an anisotropic diffusion PDE and performs an interpolation that is reversible. The interpolation solution is restricted to the subspace of functions that can recover the discrete input image, after an appropriate smoothing and sampling. The proposed nonlinear diffusion flow lies on this subspace while its strength and anisotropy adapt to the local variations and geometry of image structures. The derived method effectively reconstructs the real image structures and yields a satisfactory interpolation result. Compared to classic and other existing PDE-based interpolation methods, our proposed method seems to increase the accuracy of the result and to reduce the undesirable artifacts, such as blurring, ringing, block effects and edge distortion. We present extensive experimental results that demonstrate the potential of the method as applied to graylevel and color images. The authors acknowledge the financial support of the Future and Emerging Technologies (FET) programme ‘ASPI’ within the Sixth Framework Programme for Research of the European Commission, under FET-Open contract No. 021324.     

Shape from Radiological Density

In this paper we propose a strategy to solve the problem of recovering the 3-D shape of anatomical structures from single X-ray images, i.e., the problem of Shape from Radiological Density (SFRD). In order to overcome the noninvertibility of the process of image generation, we formulate a minimal set of physical assumptions that are used to constrain SFRD and to transform it into a well-posed problem. Our shape recovery strategy requires the solution of four problems: (a) linearization of the process of X-ray image generation, (b) image segmentation, (c) estimation of a map of the local thickness of each anatomical structure of interest, and (d) recovery of the 3-D shape of each structure from its boundaries and thickness map. In this paper we assume that problems (a) and (b) have already been faced, and propose a solution for problems (c) and (d). Experimental results on synthetic images, X-ray images of phantoms, and real radiograms are reported.     

基于图像扩散速度模型和纹理信息的人脸活体检测

为了解决人脸身份认证中的欺诈问题,提出了一种基于图像扩散速度模型和纹理信息的人脸活体检测算法。真实人脸和虚假人脸图像的空间结构不同,为了提取这种差异特征,该方法使用各向异性扩散增强图像的边缘信息。然后,将原始图像与扩散后图像的差值作为图像的扩散速度,并构建扩散速度模型。接着使用局部二值算法提取图像扩散速度特征并训练分类器。真实人脸图像和虚假人脸图像之间存在很多差异特征,为了进一步提高人脸活体检测算法的泛化能力,该方法同时提取人脸图像的模糊程度特征和色彩纹理特征,通过特征矩阵级联的方法将两种特征进行融合,并训练另一个分类器。最后根据分类器输出概率加权融合的结果做出判决。实验结果表明,该算法能够快速有效地检测出虚假的人脸图像。     

Image fusion enhancement of deformable human structures using a two-stage warping-deformable strategy: A content-based image retrieval consideration

In medical image registration and content-based image retrieval, the rigid transformation model is not adequate for anatomical structures that are elastic or deformable. For human structures such as abdomen, registration would involve global features such as abdominal wall as well as local target organs such as liver or spleen. A general non-rigid registration may not be sufficient to produce image matching of both global and local structures. In this study, a warping-deformable model is proposed to register images of such structures. This model uses a two-stage strategy for image registration of abdomen. In the first stage, the global-deformable transformation is used to register the global wall. The warping-transformation is used in second stage to register the liver. There is a good match of images using the proposed method (mean similarity index = 0.73545).The image matching correlation coefficients calculated from eight pairs of CT and MR images of abdomen indicates that the warping-deformable transformation gives better matching of images than those without transformation (p < 0.001, paired t-test). This study has established a model for image registration of deformable structures. This is particularly important for data mining of image content retrieval for structures which are non-rigid. The result obtained is very promising but further clinical evaluation is needed     

P2M扩散与相干增强扩散相结合的抑制噪声方法

该文讨论保边缘的去噪问题。针对P2M扩散不能有效保持线状特征,相干增强扩散易出现虚假条纹的缺点,提出了一种P2M扩散与相干增强扩散相结合的去噪方法。首先,建立了一个P2M扩散与相干增强扩散的加权组合模型。该模型在图像边缘部分侧重于相干增强扩散,其余部分则侧重于P2M扩散。然后,针对模型中存在的参数选取问题进行了分析。从公式推导出发,得到了在边缘点百分比给定的条件下,P2M扩散参数的自适应取值方法,并从应用的角度出发,得到了相干增强扩散参数的经验取值。仿真计算结果表明,与一些常用的去噪方法相比,该方法既能有效地抑制图像噪声,又能较好地保持边缘等线状特征,同时具有较高的峰值信噪比。     

Cine viewing of abdominal CT

A few studies have been reported that CT cine viewing on the CRT is superior to film-based viewing of CT images (Seltzer et al., Radiology 197 (1994) 119; Bonaldi et al., Am. J. Roentgenol. 170 (1998) 373; Tillich et al., Am. J. Roentgenol. 169 (1997) 1611). The purpose of our study is to know how to use cine viewing of abdominal CT. Thirty CT studies on the abdomen with both precontrast and postcontrast images were examined. The suitable rate of cine viewing ranged from 1 to 6 frames per second according to the size, the contrast and the complexity of the anatomical structures, and the slice thickness. For small or complex structures, checking each image might be required to know the full detail of them. Positional sorting among multiphase images, which is followed by consecutive display of a precontrast image, postcontrast early and late phase images at one position and so on, is useful to see the dynamic pattern of enhancement of the anatomical structures. However, there was no significant difference between cine viewing and film-based viewing concerning both the detectability of the anatomical structures and the conspicuity of enhancement of the liver and the pancreas, so that cine viewing might be an alternative to film-based viewing for CT diagnosis of the abdomen.     

基于异性扩散-中值滤波的超声医学图像去噪方法

针对超声图像存在一种特殊的斑点噪声,使图像边界与细节变得模糊而严重影响图像质量的问题,提出了一种新的去除医学图像斑点噪声的方法,它利用中值滤波和各向异性扩散相结合,不仅可以有效地去除噪声而且很好地保持了边缘、局部细节信息.此外,该方法在扩散过程中,梯度阈值选取的不同对图像结果影响很小,这极大地提高了该算法的健壮性.实验中,通过和各向异性扩散、中值滤波等方法的比较,表明该方法具有良好的去噪效果.     

各向异性滤波算法在地震曲率属性中的应用

在曲率属性计算之前需要对图像进行去噪预处理,传统的图像滤波方法在去除噪声的同时会破坏边缘、线条、纹理等图像特征,而基于偏微分方程的P-M模型在平滑过程中会出现块效应.针对这些问题,提出了一种基于张量扩散的各向异性滤波的预处理方法.通过定义散布矩阵来获得丰富的图像局部结构信息,然后利用这些结构来控制扩散过程,以便实现图像的更好滤波.理论分析和实验结果表明,相较于一些常规的图像滤波算法,各向异性滤波得到的曲率属性效果更清晰、质量更高.     

General Adaptive Neighborhood-Based Pretopological Image Filtering

This paper introduces pretopological image filtering in the context of the General Adaptive Neighborhood Image Processing (GANIP) approach. Pretopological filters act on gray level image while satisfying some topological properties. The GANIP approach enables to get an image representation and mathematical structure for adaptive image processing and analysis. Then, the combination of pretopology and GANIP leads to efficient image operators. They enable to process images while preserving region structures without damaging image transitions. More precisely, GAN-based pretopological filters and GAN-based viscous pretopological filters are proposed in this paper. The viscous notion enables to adjust the filtering activity to the image gray levels. These adaptive filters are evaluated through several experiments highlighting their efficiency with respect to the classical operators. They are practically applied in both the biomedical and material application areas for image restoration, image background subtraction and image enhancement.     

基于双向耦合扩散的保持特征的边缘锐化和图像增强

在过去的十几年中,偏微分方程在图像增强中得到了越来越多的研究和应用.该文提出了一个保持特征的双向耦合扩散框架.这个框架沿着等照度线(边缘)的梯度方向,利用柔和的边缘判定实施反向扩散以锐化边缘;而相反地沿切线方向实施正向扩散以去除噪声和锯齿伪像.为了消除这两个相反的扩散力彼此之间的冲突,将算法分裂为一种耦合的格式,而且为了保持图像特征,利用图像的局部微分几何特征调整非线性扩散系数.实验结果显示,文中算法可以显著地提高被增强图像的视觉质量.     

模拟视觉感知系统的无参考模糊图像质量评价

为了获得与人类视觉感知一致的图像质量评价方法, 本文提出一种模拟视觉感知系统的无参考模糊图像质量评价方法. 该方法通过比较不同模糊程度的图像特征的相似度来度量图像质量. 首先, 通过对待测图像进行人工模糊, 获得不同模糊程度的图像. 然后, 通过视网膜模型提取图像的细节信息. 接着, 采用奇异值分解用来获得图像的内部结构信息. 之后, 将待测图像与其它不同模糊度图像之间的细节相似度和奇异值相似度作为度量图像模糊度的特征向量. 最后, 将这些度量特征向量输入支持向量回归模型(SVR)进行训练, 获得最终的图像质量评估模型.在常用数据库上的实验结果表明, 该方法与人眼主观视觉感知的一致性优于比较方法.     

流状线型结构图像修复算法的研究

针对流状线型结构图像修复问题,提出了一种基于张量扩散的流状结构图像修复算法,模型根据图像局部结构（局部图像结构由结构张量来度量）的纹理走向确定沿纹理方向和垂直纹理方向的扩散来修复断裂特征,并且控制沿纹理方向的扩散强度要大于沿垂直纹理方向的扩散强度,由此才能保证较好的纹理修复。为使计算结果更加准确,采用了非负性离散化和最优化旋转不变性两种数值计算方案。实验结果表明,对于有划痕或较小损坏区域的流状线型结构的纹理图像,该算法都能取得较好的修复结果。     

基于图像边缘检测的变系数误差扩散方法

误差扩散算法是一种重要的图像半色调化技术,被广泛应用于各种二值化的输出设备当中,但传统的误差扩散算法生成的半色调图像易出现"蠕虫"现象和边缘模糊。针对此问题,本文提出一种结合图像边缘检测的变系数误差扩散方法,该方法以Ostromoukhov提出的变系数误差扩散算法为基础,利用其优化后的误差扩散系数,可以较好地克服"蠕虫"现象。在此基础上,采用Sobel算子检测图像的边缘并沿边缘方向修正误差扩散滤波系数,以较好地保持原图像的边缘特性。实验结果表明,利用该方法产生的半色调图像不仅具有蓝噪声特性,而且边缘结构清晰。     

Smoothing of ultrasound images using a new selective average filter

Ultrasound images are strongly affected by speckle noise making visual and computational analysis of the structures more difficult. Usually, the interference caused by this kind of noise reduces the efficiency of extraction and interpretation of the structural features of interest. In order to overcome this problem, a new method of selective smoothing based on average filtering and the radiation intensity of the image pixels is proposed. The main idea of this new method is to identify the pixels belonging to the borders of the structures of interest in the image, and then apply a reduced smoothing to these pixels, whilst applying more intense smoothing to the remaining pixels. Experimental tests were conducted using synthetic ultrasound images with speckle noisy added and real ultrasound images from the female pelvic cavity. The new smoothing method is able to perform selective smoothing in the input images, enhancing the transitions between the different structures presented. The results achieved are promising, as the evaluation analysis performed shows that the developed method is more efficient in removing speckle noise from the ultrasound images compared to other current methods. This improvement is because it is able to adapt the filtering process according to the image contents, thus avoiding the loss of any relevant structural features in the input images.     

Fast Anisotropic Smoothing of Multi-Valued Images using Curvature-Preserving PDE's

We are interested in PDE's (Partial Differential Equations) in order to smooth multi-valued images in an anisotropic manner. Starting from a review of existing anisotropic regularization schemes based on diffusion PDE's, we point out the pros and cons of the different equations proposed in the literature. Then, we introduce a new tensor-driven PDE, regularizing images while taking the curvatures of specific integral curves into account. We show that this constraint is particularly well suited for the preservation of thin structures in an image restoration process. A direct link is made between our proposed equation and a continuous formulation of the LIC's (Line Integral Convolutions by Cabral and Leedom (1993). It leads to the design of a very fast and stable algorithm that implements our regularization method, by successive integrations of pixel values along curved integral lines. Besides, the scheme numerically performs with a sub-pixel accuracy and preserves then thin image structures better than classical finite-differences discretizations. Finally, we illustrate the efficiency of our generic curvature-preserving approach – in terms of speed and visual quality – with different comparisons and various applications requiring image smoothing : color images denoising, inpainting and image resizing by nonlinear interpolation.     

Connectivity-based local adaptive thresholding for carotid artery segmentation using MRA images

The computer algorithms for the delineation of anatomical structures and other regions of interest on the medical imagery are important component in assisting and automating specific radiological tasks. In addition, the segmentation of region is an important first step for variety image related application and visualization tasks. In this paper, we propose a fast and automated connectivity-based local adaptive thresholding (CLAT) algorithm to segment the carotid artery in sequence medical imagery. This algorithm provides the new feature that is the circumscribed quadrangle on the segmented carotid artery for region-of-interest (ROI) determination. By using the preserved connectivity between consecutive slice images, the size of the ROI is adjusted like a moving window according to the segmentation result of previous slice image. The histogram is prepared for each ROI and then smoothed by local averaging for the threshold selection. The threshold value for carotid artery segmentation is locally selected on each slice image and is adaptively determined through the sequence image. In terms of automated features and computing time, this algorithm is more effective than region growing and deformable model approaches. This algorithm is also applicable to segment the cylinder shape structures and tree-like blood vessels such as renal artery and coronary artery in the medical imagery. Experiments have been conducted on synthesized images, phantom and clinical data sets with various Gaussian noise.