颌骨重建中的图像分割和轮廓对应及分支问题

根据人体颌骨的形状特征,提出一系列有针对性的方法,在CT断层数据的基础上重建三维表面模型．研究的内容包括医学图像的分割和从轮廓线重建表面两个方面．首先,采用一种称为“非种子区域分割”方法提取物体的轮廓线;然后,建立一条反应颌骨走向的弓形“参照基准线”,充分利用轮廓线相对于它的“归一化位置”信息解决表面重建中的对应和分支问题;同时考虑相邻层的影响,使生成的表面更加光滑、自然．最后,提出“多阶凸包点对应匹配”的思想,用分层分段迭代的方法拼接三角面片．     

A Variational Approach for the Segmentation of the Left Ventricle in Cardiac Image Analysis

In this paper we propose a level set method to segment MR cardiac images. Our approach is based on a coupled propagation of two cardiac contours and integrates visual information with anatomical constraints. The visual information is expressed through a gradient vector flow-based boundary component and a region term that aims at best separating the cardiac contours/regions according to their global intensity properties. In order to deal with misleading visual support, an anatomical constraint is considered that couples the propagation of the cardiac contours according to their relative distance. The resulting motion equations are implemented using a level set approach and a fast and stable numerical approximation scheme, the Additive Operator Splitting. Encouraging experimental results are provided using real data.     

Semi-automated CT segmentation using optic flow and Fourier interpolation techniques

In radiotherapy treatment planning, tumor volumes and anatomical structures are manually contoured for dose calculation, which takes time for clinicians. This study examines the use of semi-automated segmentation of CT images. A few high curvature points are manually drawn on a CT slice. Then Fourier interpolation is used to complete the contour. Consequently, optical flow, a deformable image registration method, is used to map the original contour to other slices. This technique has been applied successfully to contour anatomical structures and tumors. The maximum difference between the mapped contours and manually drawn contours was 6 pixels, which is similar in magnitude to difference one would see in manually drawn contours by different clinicians. The technique fails when the region to contour is topologically different between two slices. A solution is recommended to manually delineate contours on a sparse subset of slices and then map in both directions to fill the remaining slices.     

Semi-automated CT segmentation using optic flow and Fourier interpolation techniques

In radiotherapy treatment planning, tumor volumes and anatomical structures are manually contoured for dose calculation, which takes time for clinicians. This study examines the use of semi-automated segmentation of CT images. A few high curvature points are manually drawn on a CT slice. Then Fourier interpolation is used to complete the contour. Consequently, optical flow, a deformable image registration method, is used to map the original contour to other slices. This technique has been applied successfully to contour anatomical structures and tumors. The maximum difference between the mapped contours and manually drawn contours was 6 pixels, which is similar in magnitude to difference one would see in manually drawn contours by different clinicians. The technique fails when the region to contour is topologically different between two slices. A solution is recommended to manually delineate contours on a sparse subset of slices and then map in both directions to fill the remaining slices.     

基于信号多尺度边缘表示的CT医学图像增强

基于信号的多尺度边缘表示理论框架，提出了一种CT医学图像增强算法。较之传统的窗位一窗宽CT医学图像增强技术，该算法具有自适应的特点，以及更好地满足CT医学图像放射科临床检查应用需求的性能。临床实验结果表明，该算法在CT医学图像增强领域有着良好的应用前景和实用价值，是一种新的CT医学图像增强方法。     

A SVM-based method to extract urban areas from DMSP-OLS and SPOT VGT data

Mapping urban areas at regional and global scales has become an urgent task because of the increasing pressures from rapid urbanization and associated environmental problems. Satellite imaging of stable anthropogenic lights from DMSP-OLS provides an accurate, economical, and straightforward way to map the global distribution of urban areas. To address problems in the thresholding methods that use empirical strategies or manual trial-and-error procedures, we proposed a support vector machine (SVM)-based region-growing algorithm to semi-automatically extract urban areas from DMSP-OLS and SPOT NDVI data. Several simple criteria were used to select SVM training sets of urban and non-urban pixels, and an iterative classification and training procedure was adopted to identify the urban pixels through region growing. The new method was validated using the extents of 25 Chinese cities, as classified by Landsat ETM+ images, and then compared with two common thresholding methods. The results showed that the SVM-based algorithm could not only achieve comparable results to the local-optimized threshold method, but also avoid its tedious trial-and-error procedure, suggesting that the new method is an easy and simple alternative for extracting urban extent from DMSP-OLS and SPOT NDVI data.     

一种肺部肿瘤CT图象序列的自动分割方法

肺部肿瘤序列图象的自动分割是计算机肺部肿瘤三维辅助诊断系统的关键技术之一,肿瘤与周围组织关系的复杂性造成分割困难.为了给医生提供准确的肺部肿瘤影像,运用纹理分析和径向基神经网络实现了肺部肿瘤CT图象序列的自动分割,并根据相邻层肿瘤图象灰度、位置的相关性,提出了一种自动获取多层肿瘤区域神经网络训练样本的阈值分割算法.该算法首先计算图象纹理统计参数,以组成特征矢量空间,然后利用自适应径向基神经网络对特征矢量进行分类来实现肿瘤序列图象的自动分割.实验结果表明,与基于灰度的区域增长法和基于梯度算子和形状算子的最优阈值的分割方法相比较,该方法不仅能充分利用肺部肿瘤序列图象的三维信息,还可最大限度地减少人工干预,且分割结果较好地表现了肿瘤形态特征,经临床医生评估,具有较好的临床指导价值.     

医学图像3维重建模型的虚拟剖切算法

对医学图像体数据及重构几何模型进行虚拟剖切,可以方便地看到内部的组织,便于观察和诊断,可用于医疗放射治疗规划.针对医学图像重建的表面几何模型,提出了对模型进行平面剖切、立体开窗及任意交互切割的算法.平面剖切和开窗是用剖切面或剖切体对重建模型施以剖切,在剖切面上生成边序列及顶点序列;由此边序列和顶点序列生成封闭的边界轮廓,确定各轮廓的包含关系;对封闭轮廓包围的截面区域进行Delaunay三角剖分,得到完整的剖切后的表面模型.任意交互切割过程是交互生成切割路径,确定切割边界,并沿切割边界对表面模型进行切割.实验结果证明了本文算法的有效性.采用本文算法可得到良好的虚拟剖切效果.     

脑血肿CT图像分割的改进快速行进方法

为解决脑血肿CT图像的手动分割和半自动分割方法在手术计划制定过程中存在分割速度无法满足临床时间要求的问题,对快速行进(fast marching,FM)方法进行改进,提出一种三维全自动脑血肿CT图像分割方法。利用阈值处理自动获取种子点集,解决手动设置种子点的问题,实现全自动;提取感兴趣区域减少计算量;通过感兴趣区域金字塔的迭代分割进一步减少FM方法的演化计算量,使分割结果快速向边界收敛。实验结果表明,该方法能够自动、准确并高效地分割脑血肿CT图像。     

Development and applications of a software for Functional Image Registration (FIRE)

Image registration with anatomical modalities, such as CT and MRI, facilitates the anatomical identification and localization in the interpretation of nuclear medicine images that lack anatomical information. The implementation of Functional Image Registration (FIRE), an operating system (OS) and platform independent multimodal image registration software is reported. In order to register the images without an operator's interaction, several automatic algorithms were implemented. These include principal axes matching and maximization of the mutual information methods. The user interface was designed to support the manual registration of the images. Fused images were composed by overlaying one image with the other one transparently, in which the opacity of the overlaid image was interactively controlled. FIRE was successfully applied to many clinical cases for which automatic and/or manual registration was required. An OS and platform independent program for image registration developed in this study will be useful for the clinical application of image registration techniques.     

A Flexible,Generic Model for Anatomic Shape: Application to Interactive Two-Dimensional Medical image Segmentation and Matching

A representation called a radial contour model (RCM) is described for two-dimensional anatomic shapes. The model, which is a type of a geometric constraint network (GCN), is both flexible, in that it can deform to fit a particular instance of an anatomic shape, and generic, in that it captures all examples of a particular anatomic shape class. The model is implemented in a program, called SCANNER (version 0.7), for interactive model-based two-dimensional image segmentation and matching. Use of the model allows the segmenter to direct the search for edges in the image, and to fill in edges where none are present. Evaluations were done using models of 15 cross-sectional shapes appearing on CT images from 16 patients. Results from 480 trials show that the model-based approach reduces segmentation time by nearly a factor of 3 over manual methods, and correctly classifies 72.9% of the contours. The results not only suggest that the RCM will be useful for several current medical image segmentation tasks, but also support the hypothesis that geometric constraint networks are a viable approach to anatomic shape representation.     

Automated Segmentation of Left Ventricle Using Local and Global Intensity Based Active Contour and Dynamic Programming

The aim of this work is to develop an improved region based active contour and dynamic programming based method for accurate segmentation of left ventricle (LV) from multi-slice cine short axis cardiac magnetic resonance (MR) images. Intensity inhomogeneity and weak object boundaries present in MR images hinder the segmentation accuracy. The proposed active contour model driven by a local Gaussian distribution fitting (LGDF) energy and an auxiliary global intensity fitting energy improves the accuracy of endocardial boundary detection. The weightage of the global energy fitting term is dynamically adjusted using a spatially varying weight function. Dynamic programming scheme proposed for the segmentation of epicardium considers the myocardium probability map and a distance weighted edge map in the cost matrix. Radial distance weighted technique and conical geometry are employed for segmenting the basal slices with left ventricle outflow tract (LVOT) and most apical slices. The proposed method is validated on a public dataset comprising 45 subjects from medical image computing and computer assisted interventions (MICCAI) 2009 segmentation challenge. The average percentage of good endocardial and epicardial contours detected is about 99%, average perpendicular distance of the detected good contours from the manual reference contours is 1.95 mm, and the dice similarity coefficient between the detected contours and the reference contours is 0.91. Correlation coefficient and the coefficient of determination between the ejection fraction measurements from manual segmentation and the automated method are respectively 0.978 1 and 0.956 7, for LV mass these values are 0.924 9 and 0.855 4. Statistical analysis of the results reveals a good agreement between the clinical parameters determined manually and those estimated using the automated method.     

Using anatomical knowledge expressed as fuzzy constraints to segment the heart in CT images

Segmenting the heart in medical images is a challenging and important task for many applications. In particular, segmenting the heart in CT images is very useful for cardiology and oncological applications such as radiotherapy. Although the majority of methods in the literature are designed for ventricle segmentation, there is a real interest in segmenting the heart as a whole in this modality. In this paper, we address this problem and propose an automatic and robust method, based on anatomical knowledge about the heart, in particular its position with respect to the lungs. This knowledge is represented in a fuzzy formalism and it is used both to define a region of interest and to drive the evolution of a deformable model in order to segment the heart inside this region. The proposed method has been applied on non-contrast CT images and the obtained results have been compared to manual segmentations of the heart, showing the good accuracy and high robustness of our approach.     

数字图像离散线性变换的基于边过程的方法

本文提出了一个对数字图像进行离散线性变换的基于边过程的方法，这个方法是先做基于边过程的围线追踪；再对由边过程组成的围线进行变换，这时变换能高效率地实现；最后对变换所得区域进行填充。这一方法能适用于任意的线性变换，能使变换结果更为真实和准确，能简洁快速地实现。视围线围出区域为区域域，在此意义下，本文方法可以看做是点辖方法的一种推广。     

Interactive 3D medical image segmentation with energy-minimizing implicit functions

We present an interactive segmentation method for 3D medical images that reconstructs the surface of an object using energy-minimizing, smooth, implicit functions. This reconstruction problem is called variational interpolation. For an intuitive segmentation of medical images, variational interpolation can be based on a set of user-drawn, planar contours that can be arbitrarily oriented in 3D space. This also allows an easy integration of the algorithm into the common manual segmentation workflow, where objects are segmented by drawing contours around them on each slice of a 3D image.Because variational interpolation is computationally expensive, we show how to speed up the algorithm to achieve almost real-time calculation times while preserving the overall segmentation quality. Moreover, we show how to improve the robustness of the algorithm by transforming it from an interpolation to an approximation problem and we discuss a local interpolation scheme.A first evaluation of our algorithm by two experienced radiology technicians on 15 liver metastases and 1 liver has shown that the segmentation times can be reduced by a factor of about 2 compared to a slice-wise manual segmentation and only about one fourth of the contours are necessary compared to the number of contours necessary for a manual segmentation.     

Automatic construction of 2D shape models

A procedure for automated 2D shape model design is presented. The system is given a set of training example shapes defined by contour point coordinates. The shapes are automatically aligned using Procrustes analysis and clustered to obtain cluster prototypes (typical objects) and statistical information about intracluster shape variation. One difference from previous methods is that the training set is first automatically clustered and shapes considered to be outliers are discarded. In this way, cluster prototypes are not distorted by outliers. A second difference is in the manner in which registered sets of points are extracted from each shape contour. We propose a flexible point matching technique that takes into account both pose/scale differences and nonlinear shape differences. The matching method is independent of the objects' initial relative position/scale and does not require any manually tuned parameters. Our shape model design method was used to learn 11 different shapes from contours that were manually traced in MR brain images. The resulting model was then employed to segment several MR brain images that were not included in the shape-training set. A quantitative analysis of our shape registration approach, within the main cluster of each structure, demonstrated results that compare very well to those achieved by manual registration; achieving an average registration error of about 1 pixel. Our approach can serve as a fully automated substitute to the tedious and time-consuming manual 2D shape registration and analysis