基于数学形态学的CT图像肝脏肿瘤提取研究

目的 肝脏肿瘤的提取是肝脏三维可视化、手术规划和模拟的基础,而当前肿瘤分割存在干预过多和分割效果不佳的问题.方法 本文通过对腹部CT图像进行高斯平滑以去除图像噪声和细密纹理,计算出图像的形态学梯度并用高、低帽变换进行增强,再根据用户选择点计算内部和外部标记符,然后基于控制标记符的分水岭算法分割图像,提取出腹部CT图像中的病变组织.结果 实验结果表明,该算法能够在较少的人工干预下快速分割出肝脏病变组织.结论 该算法实现了腹部CT图像中肝脏病变组织的提取.     

基于MATLAB的CT肝脏肿瘤图像分割

目的 将CT图像中的肝脏肿瘤部分进行准确分割.方法 利用MATLAB平台对CT肝脏肿瘤图像进行预处理,并结合灰度转换、二值化处理、反色处理、形态学处理、区域生长法对病灶区域进行分割.结果 组合分割法能够发挥简单、快速、适合小病灶区域的分割特点,实现了肝脏肿瘤组织的分割,分割效果理想.结论 该方法用于肝部肿瘤的分割具有一定的有效性,但对于与周围粘连较多的肿瘤的分割还有局限性.     

Segmentation for CT images of liver neoplasms based on MATLAB

目的 将CT图像中的肝脏肿瘤部分进行准确分割.方法 利用MATLAB平台对CT肝脏肿瘤图像进行预处理,并结合灰度转换、二值化处理、反色处理、形态学处理、区域生长法对病灶区域进行分割.结果 组合分割法能够发挥简单、快速、适合小病灶区域的分割特点,实现了肝脏肿瘤组织的分割,分割效果理想.结论 该方法用于肝部肿瘤的分割具有一定的有效性,但对于与周围粘连较多的肿瘤的分割还有局限性.     

一种基于CT图像的肝肿瘤组合分割法

目的研究如何对医学CT图像中的肝肿瘤进行分割。方法采用一种结合二值化处理、区域生长法、边界分割法的组合分割方法,对CT图像进行降噪和边缘锐化。对26张不同的CT图像上的肝肿瘤作了分割操作,验证该方法的可行性。结果该方法能将所研究的肝肿瘤从CT图像中准确地分割出来,同时对于轮廓清晰、与周围组织粘连少的肝部CT肿瘤图像分割较好。大部分能够实现理想分割,但仍有少数的分割效果与实际相差较远。结论该方法对一般的肝肿瘤的分割具有一定的有效性,但对其他肿瘤,特别是与周围粘连较多的肿瘤的分割还具有一定的局限性。     

基于CT图像对人体肝副裂的观察研究

目的 总结分析肝副裂的CT影像,以期提高对肝副裂的认识,避免误诊及手术损害。方法回顾性分析2020年9月至12月东莞康华医院1449例上腹部CT,观察肝副裂的检出率及影像表现,记录其分布及形态。结果 1449例CT中共检出205例具有肝副裂,其CT发现率为14.1%(205/1449),其中男性125例（61%）,女性80例（39%）,不同性别肝副裂检出率无统计学差异（χ2=0.592,P>0.05）;所检出肝副裂分布以肝右前叶下段发生率最高（55%）;随着年龄的增长,肝副裂的数目、长度、宽度随之增加,且具有统计学差异（P<0.05）;肝副裂CT影像可表现为波浪型（24例）、膈褶型（5例）、楔型（107例）、沟型（68例）及不规则型（1例）。结论 肝副裂在人群中分布广泛,多排螺旋CT可有效地观察到肝副裂的存在,了解其形态、分布特点,有助于减少对肝副裂的误诊。     

基于自适应标记分水岭算法的肝脏CT图像自动分割

目的为减少人工交互提出了基于自适应标记分水岭的CT系列图像肝脏区域自动分割算法。方法首先对图像进行形态学重构运算以平滑图像,然后计算多尺度形态学梯度,同时提出利用梯度图像非零的局部极小值点的均值进行自适应标记提取,以避免分水岭的过分割和欠分割,再结合肝脏为最大的实质性脏器和相邻图像的相似性实现CT系列图像的肝区自动分割。结果该算法能自动、快速地提取CT系列图像中的肝脏区域。结论分水岭算法能准确定位区域的边缘,通过选择合适的阈值对梯度图像进行标记以抑制分水岭的过分割,实现医学图像中感兴趣区域的自动分割。     

基于2.5D级联卷积神经网络的CT图像胰腺分割方法

目的:由于胰腺体积小、形态个体差异性大，影像上的准确分割较为困难。本文提出一种基于2.5D级联卷积神经网络的CT图像胰腺分割方法。方法:实验中使用的数据为NIH胰腺分割公开数据集，共包含82例腹部CT图像，随机选取其中56、9、17例分别作为训练集、验证集和测试集；训练过程中使用旋转、拉伸、平移、裁剪等操作对数据进行扩增。实验中提出一种用于胰腺分割的、结合概率图的2.5D级联深度监督UNet，即CSNet（Cascading deep Supervision UNet）。该网络由3个部分组成:第1部分基于UNet，输入连续5层图像，输出中间3层对应的粗分割图像，设置适当的阈值，使其变成二值的粗分割结果；第2部分将第1层、第3层的粗分割结果与中间层的原始图像相结合，输入另一个深度监督UNet网络，得到中间层的精细分割；第3部分将第1部分网络输出的中间层的粗分割概率图与第2部分网络输出的细分割概率图通过1×1卷积进行概率融合得到最终的输出结果。3个子网络同时进行训练，对应的能量函数联合优化，从而得到更精准的分割结果。最后，使用DSC对分割结果进行评估。结果:在独立测试集上，CSNet实现了（83.74±5.27）%的DSC值。结论:CSNet可以准确分割出CT图像上的胰腺区域。     

基于双源CT图像的冠状动脉分割

背景：在计算机辅助下,从双源CT图像中把三维冠状动脉分割出来能为其定量评价提供基础。但冠状动脉的三维形态复杂多变,且其管径细小,因而实现冠状动脉的高精度分割是一项有挑战性的课题。 目的：解决冠状动脉难以实现高精度分割的问题。 方法：采用三步数据处理策略实现冠状动脉分割。先采用阈值方法对三维双源CT图像进行预分割;然后,采用交互式的策略分割出与主动脉相连的左、右冠状动脉始端;最后,根据冠状动脉始端的位置,利用形态学方法和三维断层图像相邻层间的关系分割出三维冠状动脉。 结果与结论：提出的基于形态学与断层图像层间关系的分割方法能较精确地从双源CT图像中分割出左、右冠状动脉,说明该方法适用于三维冠状动脉的分割。     

基于VTK的肝组织CT断层图像三维重建

目的利用可视化开发包（VTK）将计算机断层扫描（cT）图像合成三维立体图像,以利于更好地制定肝癌患者的治疗方案。方法建立基于VC6．0平台上的VTK可视开发环境,获取患者的DICOM格式的CT图像,采用VTKfilter滤波器和拉普拉斯锐化模板对图像进行滤波、增强等预处理,利用光线投影Ray—Casting算法对图像进行三维重建。结果该方法可生成效果较好的三维图像,其分辨率较高,图像重建速度快,且可进行切割、旋转、缩放等功能。结论基于VTK的光线投影算法可用于对人肝组织的三维重建,有利于医生直观地观察病变组织的形状和位置,为诊断和治疗提供依据。     

一种基于相位相关法及数学形态学方法的眼底血管图像自动拼

Precise mosaic techniques are essential for quantitative evaluation of retinal images to make early detection of fundus anomalies feasible. Opening of a gray-scale image by a gray-scale structuring element(SE) can generate a background image. Image mosaic was achieved by subtraction this background image from the original image and then applying a phase correlation method to find translation difference. Because of the accuracy characteristics of the phase correlation method and the speed of the FFT hardware, this new algorithm can work very fast and accurately, compared to conventional techniques. The method was also characterized by an outstanding robustness against correlated noise and disturbances, such as those encountered with nonuniform illumination. The results confirm the robustness of the chosen approach.     

Efficient Denoising Technique for CT images to Enhance Brain Hemorrhage Segmentation

This paper presents an adaptive denoising approach aiming to improve the visibility and detectability of hemorrhage from brain computed tomography (CT) images. The suggested approach fuses the images denoised by total variation (TV) method, denoised by curvelet-based method, and edge information extracted from the noise residue of TV method. The edge information is extracted from the noise residue of TV method by processing it through curvelet transform. The visual interpretation shows that the proposed approach not only reduces the staircase effect caused by total variation method but also reduces visual distortion induced by curvelet transform in the homogeneous areas of the CT images. The denoising abilities of the proposed method are further evaluated by segmenting the hemorrhagic brain area using region-growing method. The sensitivity, specificity, Jaccard index, and Dice coefficients were calculated for different noise levels. The comparative results show that the significant improvement has yielded in the brain hemorrhage detection from CT images after denoising it with the proposed approach.     

基于数学形态学的视网膜血管提取算法

目的 影像中血管的分割与特征提取,对疾病的早期诊断具有重要意义。针对很多视网膜血管提取算法分割精度不高的问题,提出了运用数学形态学中的高帽变换的方法对其进行检测。方法 首先,选取结构元素为“圆盘形”的形态学对图像进行高帽变换,经过高帽变换后的图像平滑了图像的背景,同时增强了血管在图像中的对比度。其次,对变换后的图像利用Otsu＇s自动分割法对图像进行阈值分割得到血管的二值图像。再次,根据血管在图像中的结构信息和几何信息,利用基于连通域度量的方法,设置连通域的“面积”和“长宽比”两个阈值,去除虚假目标。最后,为保持血管的连续性,对图像进行一次膨胀运算,可将断裂的血管连接起来,减小了实验的误差。结果 通过上述步骤实现了对血管的提取。结论 结果表明,本文算法能有效提取视网膜眼底图像的血管网络,有较强的分割精度。     

Attenuation correction of PET images with interpolated average CT for thoracic tumors

To reduce positron emission tomography (PET) and computed tomography (CT) misalignments and standardized uptake value (SUV) errors, cine average CT (CACT) has been proposed to replace helical CT (HCT) for attenuation correction (AC). A new method using interpolated average CT (IACT) for AC is introduced to further reduce radiation dose with similar image quality. Six patients were recruited in this study. The end-inspiration and -expiration phases from cine CT were used as the two original phases. Deformable image registration was used to generate the interpolated phases. The IACT was calculated by averaging the original and interpolated phases. The PET images were then reconstructed with AC using CACT, HCT and IACT, respectively. Their misalignments were compared by visual assessment, mutual information, correlation coefficient and SUV. The doses from different CT maps were analyzed. The misalignments were reduced for CACT and IACT as compared to HCT. The maximum SUV difference between the use of IACT and CACT was ～3%, and it was ～20% between the use of HCT and CACT. The estimated dose for IACT was 0.38 mSv. The radiation dose using IACT could be reduced by 85% compared to the use of CACT. IACT is a good low-dose approximation of CACT for AC.     

基于胸部CT图像的肺结节分割

目的提出一种从胸部CT图像中分割提取多种类型肺结节的算法,辅助肺癌诊断和疗效评估。方法首先由放射科医生确定种子点和目标容积区域,再根据初分割结果自动识别非肺壁粘连结节和肺壁粘连结节。然后采用多阈值结合距离变换的方法分割非肺壁粘连结节,光线投射和直线拟合分割肺壁粘连结节。最后,将算法应用于85组患者数据（232个肺结节）,并由高年资放射科医生评价分割结果的准确性。结果本文算法鲁棒性强,能准确判别肺壁粘连和非肺壁粘连结节,从而适用于孤立、血管粘连、毛玻璃和肺壁粘连结节的提取。测试的232个结节中无异常发生,且分割速度较快。经放射医生评价,平均准确率达90％。结论本文算法可以从胸部CT图像中分割提取4种类型肺结节,鲁棒性、准确性和速度均可满足实际临床需求,对肺癌筛查、诊断和疗效评估具有重要价值。     

Iterative sorting for four-dimensional CT images based on internal anatomy motion

Current four-dimensional (4D) computed tomography (CT) imaging techniques using multislice CT scanners require retrospective sorting of the reconstructed two-dimensional (2D) CT images. Most existing sorting methods depend on externally monitored breathing signals recorded by extra instruments. External signals may not always accurately capture the breathing status and may lead to severe discontinuity artifacts in the sorted CT volumes. This article describes a method to find the temporal correspondences for the free-breathing multislice CT images acquired at different table positions based on internal anatomy movement. The algorithm iteratively sorts the CT images using estimated internal motion indices. It starts from two imperfect reference volumes obtained from the unsorted CT images; then, in each iteration, thorax motion is estimated from the reference volumes and the free-breathing CT images. Based on the estimated motion, the breathing indices as well as the reference volumes are refined and fed into the next iteration. The algorithm terminates when two successive iterations attain the same sorted reference volumes. In three out of five patient studies, our method attained comparable image quality with that using external breathing signals. For the other two patient studies, where the external signals poorly reflected the internal motion, the proposed method significantly improved the sorted 4D CT volumes, albeit with greater computation time.     

Non-rigid registration between 3D ultrasound and CT images of the liver based on intensity and gradient information

In order to utilize both ultrasound (US) and computed tomography (CT) images of the liver concurrently for medical applications such as diagnosis and image-guided intervention, non-rigid registration between these two types of images is an essential step, as local deformation between US and CT images exists due to the different respiratory phases involved and due to the probe pressure that occurs in US imaging. This paper introduces a voxel-based non-rigid registration algorithm between the 3D B-mode US and CT images of the liver. In the proposed algorithm, to improve the registration accuracy, we utilize the surface information of the liver and gallbladder in addition to the information of the vessels inside the liver. For an effective correlation between US and CT images, we treat those anatomical regions separately according to their characteristics in US and CT images. Based on a novel objective function using a 3D joint histogram of the intensity and gradient information, vessel-based non-rigid registration is followed by surface-based non-rigid registration in sequence, which improves the registration accuracy. The proposed algorithm is tested for ten clinical datasets and quantitative evaluations are conducted. Experimental results show that the registration error between anatomical features of US and CT images is less than 2 mm on average, even with local deformation due to different respiratory phases and probe pressure. In addition, the lesion registration error is less than 3 mm on average with a maximum of 4.5 mm that is considered acceptable for clinical applications.     

Segmentation of bone CT images and assessment of bone structure using measures of complexity

A nondestructive and noninvasive method for numeric characterization (quantification) of the structural composition of human bone tissue has been developed and tested. In order to quantify and to compare the structural composition of bones from 2D computed tomography (CT) images acquired at different skeletal locations, a series of robust, versatile, and adjustable image segmentation and structure assessment algorithms were developed. The segmentation technique facilitates separation from cortical bone and standardizes the region of interest. The segmented images were symbol-encoded and different aspects of the bone structural composition were quantified using six different measures of complexity. These structural examinations were performed on CT images of bone specimens obtained at the distal radius, humeral mid-diaphysis, vertebral body, femoral head, femoral neck, proximal tibia, and calcaneus. In addition, the ability of the noninvasive and nondestructive measures of complexity to quantify trabecular bone structure was verified by comparing them to conventional static histomorphometry performed on human fourth lumbar vertebral bodies. Strong correlations were established between the measures of complexity and the histomorphometric parameters except for measures expressing trabecular thickness. Furthermore, the ability of the measures of complexity to predict vertebral bone strength was investigated by comparing the outcome of the complexity analysis of the CT images with the results of a biomechanical compression test of the third lumbar vertebral bodies from the same population as used for histomorphometry. A multiple regression analysis using the proposed measures including structure complexity index, structure disorder index, trabecular network index, index of a global ensemble, maximal L-block, and entropy of x-ray attenuation distribution revealed an excellent relationship (r=0.959, r2=0.92) between the measures of complexity and compressive bone strength. In conclusion, the image segmentation techniques and the assessment of bone architecture by measures of complexity have been successfully applied to analyze high-resolution peripheral quantitative computed tomography (pQCT) and CT images obtained from the distal radius, humeral mid-diaphysis, third and fourth lumbar vertebral bodies, proximal femur, proximal tibia, and calcaneus. The proposed approach is of broad interest as it can be applied for the quantification of structures and textures originating from different imaging modalities in other fields of science.     

Segmentation of pulmonary nodules in three-dimensional CT images by use of a spiral-scanning technique

Accurate segmentation of pulmonary nodules in computed tomography (CT) is an important and difficult task for computer-aided diagnosis of lung cancer. Therefore, the authors developed a novel automated method for accurate segmentation of nodules in three-dimensional (3D) CT. First, a volume of interest (VOI) was determined at the location of a nodule. To simplify nodule segmentation, the 3D VOI was transformed into a two-dimensional (2D) image by use of a key "spiral-scanning" technique, in which a number of radial lines originating from the center of the VOI spirally scanned the VOI from the "north pole" to the "south pole." The voxels scanned by the radial lines provided a transformed 2D image. Because the surface of a nodule in the 3D image became a curve in the transformed 2D image, the spiral-scanning technique considerably simplified the segmentation method and enabled reliable segmentation results to be obtained. A dynamic programming technique was employed to delineate the "optimal" outline of a nodule in the 2D image, which corresponded to the surface of the nodule in the 3D image. The optimal outline was then transformed back into 3D image space to provide the surface of the nodule. An overlap between nodule regions provided by computer and by the radiologists was employed as a performance metric for evaluating the segmentation method. The database included two Lung Imaging Database Consortium (LIDC) data sets that contained 23 and 86 CT scans, respectively, with 23 and 73 nodules that were 3 mm or larger in diameter. For the two data sets, six and four radiologists manually delineated the outlines of the nodules as reference standards in a performance evaluation for nodule segmentation. The segmentation method was trained on the first and was tested on the second LIDC data sets. The mean overlap values were 66% and 64% for the nodules in the first and second LIDC data sets, respectively, which represented a higher performance level than those of two existing segmentation methods that were also evaluated by use of the LIDC data sets. The segmentation method provided relatively reliable results for pulmonary nodule segmentation and would be useful for lung cancer quantification, detection, and diagnosis.     

基于数学形态学膨胀和腐蚀的医学磁共振图像的边缘检测

目的对腰椎磁共振图像进行椎间盘的边缘检测。方法数学形态学的基本运算是膨胀和腐蚀。通过对形态学运算的加权组合,可以构造出边缘检测的方法。分别对不同的检测方法进行了比较分析。结果发现它们各有特点。可以看出数学形态学具有很好的医学图像的边缘检测能力,可以获得图像连续的边缘,为后续的图像分割及目标识别等研究奠定了基础。结论为最终实现椎间盘的虚拟仿真外科手术方案的制定、解剖结构的测量及术后评估奠定了一个良好的基础。