Enhancing breast pectoral muscle segmentation performance by using skip connections in fully convolutional network

The precise detection and segmentation of pectoral muscle areas in mediolateral oblique (MLO) views is an essential step in the development of a computer-aided diagnosis system to access breast malignant lesions or parenchyma. The goal of this article is to develop a robust and fully automatic algorithm for pectoral muscle segmentation from mammography images. This paper presents an image enhancement approach that improves the quality of mammogram scans and a convolutional neural network-based fully convolutional network architecture enhanced with residual connections for automatic segmentation of the pectoral muscle from the MLO views of a digital mammogram. For this purpose, the model is tested and trained on three different mammogram datasets named MIAS, INBREAST, and DDSM. The ground truth labels of the pectoral muscle were identified under the supervision of experienced radiologists. For training and testing, 10-fold cross-validation was used. The proposed model was compared with baseline U-Net-based architecture. Finally, we used a postprocessing step to find the actual boundary of the pectoral muscle. Our presented architecture generated a mean Intersection over Union (IoU) of 97%, dice similarity coefficient (DSC) of 96% and 98% accuracy on testing data. The proposed architecture for pectoral muscle segmentation from the MLO views of mammogram images with high accuracy and dice score can be quickly merged with the breast tumor segmentation problem.     

Automatic COVID‐19 CT segmentation using U‐Net integrated spatial and channel attention mechanism

The coronavirus disease (COVID‐19) pandemic has led to a devastating effect on the global public health. Computed Tomography (CT) is an effective tool in the screening of COVID‐19. It is of great importance to rapidly and accurately segment COVID‐19 from CT to help diagnostic and patient monitoring. In this paper, we propose a U‐Net based segmentation network using attention mechanism. As not all the features extracted from the encoders are useful for segmentation, we propose to incorporate an attention mechanism including a spatial attention module and a channel attention module, to a U‐Net architecture to re‐weight the feature representation spatially and channel‐wise to capture rich contextual relationships for better feature representation. In addition, the focal Tversky loss is introduced to deal with small lesion segmentation. The experiment results, evaluated on a COVID‐19 CT segmentation dataset where 473 CT slices are available, demonstrate the proposed method can achieve an accurate and rapid segmentation result on COVID‐19. The method takes only 0.29 second to segment a single CT slice. The obtained Dice Score and Hausdorff Distance are 83.1% and 18.8, respectively.     

Mammographic mass segmentation using multichannel and multiscale fully convolutional networks

Breast cancer is one of the leading causes of death among women worldwide. Mammographic mass segmentation is an important task in mammogram analysis. This process, however, poses a prominent challenge considering that masses can be obscured in images and appear with irregular shapes and low image contrast. In this study, a multichannel, multiscale fully convolutional network is proposed and evaluated for mass segmentation in mammograms. To reduce the impact of surrounding unrelated structures, preprocessed images with a salient mass appearance are obtained as the second input channel of the network. Furthermore, to jointly conduct fine boundary delineation and global mass localization, we incorporate more crucial context information by learning multiscale features from different resolution levels. The performance of our segmentation approach is compared with that of several traditional and deep-learning-based methods on the popular DDSM and INbreast datasets. The evaluation indices consist of the Dice similarity coefficient, area overlap measure, area undersegmentation measure, area oversegmentation measure, and Hausdorff distance. The mean values of the Dice similarity coefficient and Hausdorff distance of our proposed segmentation method are 0.915 ± 0.031 and 6.257 ± 3.380, respectively, on DDSM and 0.918 ± 0.038 and 2.572 ± 0.956, respectively, on INbreast, which are superior to those of the existing methods. The experimental results verify that our proposed multichannel, multiscale fully convolutional network can reliably segment masses in mammograms.     

Mammographic mass detection by adaptive thresholding and region growing

We present an efficient method to detect mass lesions on digitized mammograms, which consists of breast region extraction, region partitioning, automatic seed selection, segmentation by region growing, feature extraction, and neural network classification. The method partitions the breast region into a fat region, a fatty and glandular region, and a dense region, so that different threshold values can be applied to each partitioned region during processes of the seed selection and segmentation. The mammographic masses are classified by using four features representing shape, density, and margin of the segmented regions. The method detects subtle mass lesions with various contrast ranges and can facilitate a procedure of mass detection in computer‐aided diagnosis systems. © 2001 John Wiley & Sons, Inc. Int J Imaging Syst Technol, 11, 340–346, 2000     

FBG编解码器在W-OCDMA系统中的应用

针对大容量二维组合码,提出一种可调光纤布拉格光栅(FBG)编解码器的设计方法,该设计方法利用了光纤布拉格光栅的反射和可调谐特性,所得编解码器结构简单、容易集成、变址容易。在 OOK调制输出速率为 1Gbit/s,输出功率为 1mW的 W-OCDMA 系统仿真实验中,该 FBG编解码器结构实现了正确的编解码,时间延迟 1ns,功率损耗约 1dB,并且具有良好的匹配滤波特性。该 FBG 编解码器将是适用于 W-OCDMA 系统较为理想的器件。     

A fuzzy logic‐based meningioma tumor detection in magnetic resonance brain images using CANFIS and U‐Net CNN classification

This article develops a methodology for meningioma brain tumor detection process using fuzzy logic based enhancement and co‐active adaptive neuro fuzzy inference system and U‐Net convolutional neural network classification methods. The proposed meningioma tumor detection process consists of the following stages as, enhancement, feature extraction, and classifications. The enhancement of the source brain image is done using fuzzy logic and then dual tree‐complex wavelet transform is applied to this enhanced image at different levels of scale. The features are computed from the decomposed sub band images and these features are further classified using CANFIS classification method which identifies the meningioma brain image from nonmeningioma brain image. The performance of the proposed meningioma brain tumor detection and segmentation system is analyzed in terms of sensitivity, specificity, segmentation accuracy, and Dice coefficient index with detection rate.     

Automatic cervical cancer segmentation in multimodal magnetic resonance imaging using an EfficientNet encoder in UNet++ architecture

Automatic cervical cancer segmentation in multimodal magnetic resonance imaging (MRI) is essential because tumor location and delineation can support patients' diagnosis and treatment planning. To meet this clinical demand, we present an encoder–decoder deep learning architecture which employs an EfficientNet encoder in the UNet++ architecture (E-UNet++). EfficientNet helps in effectively encoding multiscale image features. The nested decoders with skip connections aggregate multiscale features from low-level to high-level, which helps in detecting fine-grained details. A cohort of 228 cervical cancer patients with multimodal MRI sequences, including T2-weighted imaging, diffusion-weighted imaging, apparent diffusion coefficient imaging, contrast enhancement T1-weighted imaging, and dynamic contrast-enhanced imaging (DCE), has been explored. Evaluations are performed by considering either single or multimodal MRI with standard segmentation quantitative metrics: dice similarity coefficient (DSC), intersection over union (IOU), and 95% Hausdorff distance (HD). Our results show that the E-UNet++ model can achieve DSC values of 0.681–0.786, IOU values of 0.558–0.678, and 95% HD values of 3.779–7.411 pixels in different single sequences. Meanwhile, it provides DSC values of 0.644 and 0.687 on three DCE subsequences and all MRI sequences together. Our designed model is superior to other comparative models, which shows the potential to be used as an artificial intelligence tool for cervical cancer segmentation in multimodal MRI.     

A locally adaptive region growing algorithm for vascular segmentation

To segment vascular structures in 3‐D CTA/MRA images, this article presents a new region growing algorithm based on local cube tracking. In the proposed algorithm, a small local cube is segmented to detect a vessel segment, and the following local cube(s) is determined based on the segmentation result. This procedure is repeated until the segmentation is completed. By confining the segmentation inside each local cube, a robust result can be obtained even in a tubular structure of steadily changing intensity. For segmentation, a locally adaptive and competitive region growing scheme is adopted to obtain well‐defined vessel boundaries. It should be emphasized that the proposed algorithm can detect all branches with practically acceptable computational complexity. In addition, its segmentation result is represented as a tree structure having many branches so that a user may easily correct the result branch‐by‐branch, if necessary. Experimental results from real images prove that the proposed algorithm produces prospective vessel segmentation results for 3‐D CTA/MRA images and segments vessels of various sizes well, including stenoses and aneurysms. © 2003 Wiley Periodicals, Inc. Int J Imaging Syst Technol 13, 208–214, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ima.10059     

基于双分支多尺度融合网络的毫米波SAR图像多目标语义分割方法

在毫米波合成孔径雷达(SAR)安检成像违禁品的检测与识别中,存在着目标尺寸过小、目标被部分遮挡和多目标之间重叠等复杂情况,不利于违禁品的准确识别。针对这些问题,提出了一种基于双分支多尺度融合网络(DBMFnet)的违禁品检测方法。该网络使用Encoder-Decoder的结构,在Encoder阶段,提出一种双分支并行特征提取网络(DBPFEN)来增强特征提取;在Decoder阶段,提出一种多尺度融合模块(MSFM)来提高对目标的检测能力。实验结果表明,该方法的均交并比(mIoU)均优于现有的语义分割方法,降低了漏检与错检率。

     

基于注意力机制的轻量级RGB-D图像语义分割网络

目的针对卷积神经网络在RGB-D(彩色-深度)图像中进行语义分割任务时模型参数量大且分割精度不高的问题,提出一种融合高效通道注意力机制的轻量级语义分割网络。方法文中网络基于RefineNet,利用深度可分离卷积(Depthwiseseparableconvolution)来轻量化网络模型,并在编码网络和解码网络中分别融合高效的通道注意力机制。首先RGB-D图像通过带有通道注意力机制的编码器网络,分别对RGB图像和深度图像进行特征提取;然后经过融合模块将2种特征进行多维度融合;最后融合特征经过轻量化的解码器网络得到分割结果,并与RefineNet等6种网络的分割结果进行对比分析。结果对提出的算法在语义分割网络常用公开数据集上进行了实验,实验结果显示文中网络模型参数为90.41 MB,且平均交并比(mIoU)比RefineNet网络提高了1.7%,达到了45.3%。结论实验结果表明,文中网络在参数量大幅减少的情况下还能提高了语义分割精度。     

Optimum linear decoding of vector quantisation transmitted over a CDMA channel

The problem of joint source-channel and multiuser decoding for code division multiple access channels is considered. The block source-channel encoder is defined by a vector quantiser (VQ). The jointly optimum solution to such a problem has been considered before, but its extremely high complexity makes it impractical for systems with medium to large number of users and/or medium to large size of VQ codebook. Instead, the optimum linear decoder with a much lower complexity that minimises the mean-squared error is introduced. The optimum linear decoder is soft in the sense that it utilises all the soft information available at the receiver. Analytical and simulation results show that at low channel signal-to-noise ratio region, the proposed decoder's performance is almost the same as that of the jointly optimum decoder and significantly better than that of the tandem approaches that use separate multiuser detection and table-lookup decoding.     

Segmentation for brain magnetic resonance images using dual‐tree complex wavelet transform and spatial constrained self‐organizing tree map

In brain MR images, the noise and low‐contrast significantly deteriorate the segmentation results. In this paper, we introduce a novel application of dual‐tree complex wavelet transform (DT‐CWT), and propose an automatic unsupervised segmentation method integrating DT‐CWT with self‐organizing map for brain MR images. First, a multidimensional feature vector is constructed based on the intensity, low‐frequency subband of DT‐CWT, and spatial position information. Then, a spatial constrained self‐organizing tree map (SCSOTM) is presented as the segmentation system. It adaptively captures the complicated spatial layout of the individual tissues, and overcomes the problem of overlapping gray‐scale intensities for different tissues. SCSOTM applies a dual‐thresholding method for automatic growing of the tree map, which uses the information from the high‐frequency subbands of DT‐CWT. The proposed method is validated by extensive experiments using both simulated and real T1‐weighted MR images, and compared with the state‐of‐the‐art algorithms. © 2014 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 24, 208–214, 2014     

融合色彩分割和加权非参数变换的立体匹配

针对传统非参数变换的局限性,提出了一种立体匹配算法。利用参考图像的色彩分割信息获得基于任意形状和大小支持区域的匹配代价,并在相似色彩区域内计算加权非参数变换匹配代价,再将这两种匹配代价融合构成联合匹配代价,通过局部优化方法获得稠密视差图。实验结果表明算法提高了低纹理区域、遮挡区域和不连续区域的匹配精度,并对幅度失真具有鲁棒性。     

Fuzzy generalized fast marching method for 3D segmentation of brain structures

The aim of this work is to develop a new model for segmentation of brain structures in medical brain MR images. Brain segmentation is a challenging task due to the complex anatomical structure of brain structures as well as intensity nonuniformity, partial volume effects and noise. Generally the structures of interest are of relatively complicated size and have significant shape variations, the structures boundaries may be blurry or even missing, and the surrounding background is full of irrelevant edges. Segmentation methods based on fuzzy models have been developed to overcome the uncertainty caused by these effects. In this study, we propose a robust and accurate brain structures segmentation method based on a combination of fuzzy model and deformable model. Our method breaks up into two great parts. Initially, a preliminary stage allows to construct the various information maps, in particular a fuzzy map, used as a principal information source, constructed using the Fuzzy C‐means method (FCM). Then, a deformable model implemented with the generalized fast marching method (GFMM), evolves toward the structure to be segmented, under the action of a normal force defined from these information maps. In this sense, we used a powerful evolution function based on a fuzzy model, adapted for brain structures. Two extensions of our general method are presented in this work. The first extension concerns the addition of an edge map to the fuzzy model and the use of some rules adapted to the segmentation process. The second extension consists of the use of several models evolving simultaneously to segment several structures. Extensive experiments are conducted on both simulated and real brain MRI datasets. Our proposed approach shows promising and achieves significant improvements with respect to several state‐of‐the‐art methods and with the three practical segmentation techniques widely used in neuroimaging studies, namely SPM, FSL, and Freesurfer.     

Fast unsupervised texture segmentation using Texel similarity map

Unsupervised texture segmentation is a challenging topic in computer vision. It is difficult to obtain boundaries of texture regions automatically in real-time, especially for cluttered images. This paper presents a new fast unsupervised texture segmentation method. First, the Texel similarity map (TSM) is used to compare the changes of intensity and gray level of neighboring pixels to determine whether they are identical. Then, a scheme called multiple directions integral images (MDII) is proposed to quickly evaluate the TSM. With the aid of MDII, one pixel’s similarity value can be computed in constant time. Finally, the proposed segmentation method is tested on both artificial texture and natural images. Experimental results show that the proposed method performs well on a wide range of images, and outperforms state-of-the-art method on segmentation speed.     

A fully automated hybrid methodology using Cuckoo‐based fuzzy clustering technique for magnetic resonance brain image segmentation

This article aims at developing an automated hybrid algorithm using Cuckoo Based Search (CBS) and interval type‐2 fuzzy based clustering, so as to exhibit efficient magnetic resonance (MR) brain image segmentation. An automatic MR brain image segmentation facilitates and enables a radiologist to have a brief review and easy analysis of complicated tumor regions of imprecise gray level regions with minimal user interface. The tumor region having severe intensity variations and suffering from poor boundaries are to be detected by the proposed hybrid technique that could ease the process of clinical diagnosis and this tends to be the core subject of this article. The ability of the proposed technique is compared using standard comparison parameters such as mean squared error, peak signal to noise ratio, computational time, Dice Overlap Index, and Jaccard T animoto C oefficient Index. The proposed CBS combined with interval type‐2 fuzzy based clustering produces a sensitivity of 0.7143 and specificity of 0.9375, which are far better than the conventional techniques such as kernel based, entropy based, graph‐cut based, and self‐organizing maps based clustering. Appreciable segmentation results of tumor region that enhances clinical diagnosis is made available through this article and two of the radiologists who have hands on experience in the field of radiology have extended their support in validating the efficiency of the proposed methodology and have given their consent in utilizing the proposed methodology in the processes of clinical oncology.     

Slope of the Out‐of‐Plane Deflection Measurement of Composite Sandwich Plate With a Fully Potted Insert by Using Digital Phase‐Shifting Shearography

Abstract: A digital shearographic (DS) technique is a tool well‐suited for precision strain measurement and is a non‐destructive testing technology. But the fringe patterns of DS have not so high sensitivity due to the limitation of CCD camera resolution. Therefore a phase‐shifting technique is incorporated into DS and demonstrated to yield fringe patterns with good quality. The main purpose of this study is to set‐up a measuring system of digital phase‐shifting shearography (DPSS) to measure slope of the out‐of‐plane deflection of sandwich plate with a fully potted insert. The present system includes piezoelectric transducer, servo controller, Michelson shearing mechanism, image processing system and loading system. The four‐step phase shifting method is used to obtain phase map and then the phase expansion is proceeded by Macy algorithm to obtain the slope of the out‐of‐plane deflection of honeycomb sandwich plate with insert/potting material. Finally, comparing the slope of the out‐of‐plane deflection of DPSS with that of DS shows 2 to 7% difference of slope of deflection exhibiting.     

引入双编码器模型的OCT视网膜图像分割

OCT视网膜图像中存在着噪声和散斑,单一的提取空间特征往往容易遗漏一些重要信息,导致不能准确地分割目标区域。而OCT图像本身存在光谱频域特征,针对OCT图像的频域特征,本文基于U-Net和快速傅立叶卷积提出一种新的双编码器模型以提高对OCT图像视网膜层、液体的分割性能,提出的频域编码器可以提取图像频域信息并通过快速傅里叶卷积转换为空间信息,将很好地弥补单一空间编码器遗漏特征信息的不足。经过与其他经典模型的对比和消融实验,结果表明,随着频域编码器的添加,该模型能有效提升对视网膜层和液体的分割性能,平均Dice系数和mIoU相较于U-Net均提高2%,相较于ReLayNet分别提高8%和4%,其中对液体的分割提升尤为明显,相较于U-Net 模型Dice系数提高了10%。

     

Physics-integrated deep learning for uncertainty quantification and reliability estimation of nonlinear dynamical systems

Assumptions and approximations made while analyzing any physical system induce modeling uncertainty, which, if left unchecked, can result in the erroneous analysis of the system under consideration. Additionally, the discrepancy in the exact knowledge of system parameters can further result in deviation from the ground truth. This paper explores Physics-integrated Variational Auto-Encoder (PVAE) to account for modeling and parametric uncertainties in partially known nonlinear dynamical systems. The PVAE under consideration has three main parts: encoder, latent space, and decoder. The complete PVAE architecture is employed during the training stage of the machine learning model, while only the decoder is used to make the final predictions. The encoder determines the correct parameter values for the known part of the model (in the form of a known ODE). The decoder is augmented with an ODE solver that solves the known part of the system and the estimated discrepancy together to reconstruct the measurements. To test the efficacy of the PVAE architecture, three case studies are carried out, each presenting unique challenges. The probability density functions obtained for the various systems’ responses demonstrate the efficacy of the PVAE architecture. Furthermore, reliability analysis has been carried out, and the results produced have been compared against those obtained from a multi-layered, densely connected forward neural network.     

Efficient image segmentation based on deep learning for mineral image classification

Mineral image segmentation plays a vital role in the realization of machine vision based intelligent ore sorting equipment. However, the existing image segmentation methods still cannot effectively solve the problem of adhesion and overlap between mineral particles, and the segmentation performance of small and irregular particles still needs to be improved. To overcome these bottlenecks, we propose a deep learning based image segmentation method to segment the key areas in mineral images using morphological transformation to process mineral image masks. This investigation explores four aspects of the deep learning-based mineral image segmentation model, including backbone selection, module configuration, loss function construction, and its application in mineral image classification. Specifically, referring to the designs of U-Net, FCN, Seg Net, PSP Net, and DeepLab Net, this experiment uses different backbones as Encoder to building ten mineral image segmentation models with different layers, structures, and sampling methods. Simultaneously, we propose a new loss function suitable for mineral image segmentation and compare CNNs-based segmentation models' training performance under different loss functions. The experiment results show that the proposed mineral image segmentation has excellent segmentation performance, effectively solves adhesion and overlap between adjacent particles without affecting the classification accuracy. By using the Mobile Net as backbone, the PSP Net and DeepLab can achieve a high segmentation performance in mineral image segmentation tasks, and the 15 × 15 is the most suitable size for erosion element structure to process the mask images of the segmentation models.