Brain tumor segmentation in multi‐spectral MRI using convolutional neural networks (CNN)

A tumor could be found in any area of the brain and could be of any size, shape, and contrast. There may exist multiple tumors of different types in a human brain at the same time. Accurate tumor area segmentation is considered primary step for treatment of brain tumors. Deep Learning is a set of promising techniques that could provide better results as compared to nondeep learning techniques for segmenting timorous part inside a brain. This article presents a deep convolutional neural network (CNN) to segment brain tumors in MRIs. The proposed network uses BRATS segmentation challenge dataset which is composed of images obtained through four different modalities. Accordingly, we present an extended version of existing network to solve segmentation problem. The network architecture consists of multiple neural network layers connected in sequential order with the feeding of Convolutional feature maps at the peer level. Experimental results on BRATS 2015 benchmark data thus show the usability of the proposed approach and its superiority over the other approaches in this area of research.     

Feature enhancement framework for brain tumor segmentation and classification

Automatic medical image analysis is one of the key tasks being used by the medical community for disease diagnosis and treatment planning. Statistical methods are the major algorithms used and consist of few steps including preprocessing, feature extraction, segmentation, and classification. Performance of such statistical methods is an important factor for their successful adaptation. The results of these algorithms depend on the quality of images fed to the processing pipeline: better the images, higher the results. Preprocessing is the pipeline phase that attempts to improve the quality of images before applying the chosen statistical method. In this work, popular preprocessing techniques are investigated from different perspectives where these preprocessing techniques are grouped into three main categories: noise removal, contrast enhancement, and edge detection. All possible combinations of these techniques are formed and applied on different image sets which are then passed to a predefined pipeline of feature extraction, segmentation, and classification. Classification results are calculated using three different measures: accuracy, sensitivity, and specificity while segmentation results are calculated using dice similarity score. Statistics of five high scoring combinations are reported for each data set. Experimental results show that application of proper preprocessing techniques could improve the classification and segmentation results to a greater extent. However, the combinations of these techniques depend on the characteristics and type of data set used.     

基于标签预留Softmax算法的机器视觉检测鉴别语义分割迁移学习技术

面向机器视觉检测鉴别的语义分割卷积神经网络(Convolutional Neural Network,CNN)模型能识别、测量被测对象的零部件、尺寸等特征,针对机器视觉检测鉴别增加识别零部件或关键部位的需求,以及进一步迁移学习会损失CNN模型部分权值的问题,提出一种基于标签预留Softmax算法的语义分割迁移学习技术....     

多阶段边界参考网络的动作分割

针对现有动作分割算法中过分割问题导致预测错误、造成分割质量下降的现象,提出一种可调视频动作边界信息作为参考的多阶段参考网络,在基于多阶段时间卷积网络的主干网络中,为每个阶段独立引入视频动作边界信息作为参考.各阶段使用相同的边界信息会使模型固化,为使主干网络能够调整参与各阶段输出计算的边界值,对不同样本区分处理,提出多层...     

基于特征融合度量学习的高压断路器机械故障诊断

数据驱动的深度学习方法在高压断路器机械故障诊断中取得了一定的成效,然而这些方法实现优异性能的前提是可获取海量训练样本,在现场数据匮乏场景下其诊断性能明显下降。为此,提出了一种新颖的特征融合度量学习模型用于现场小样本高压断路器机械故障诊断。首先构建了特征融合卷积神经网络,有效提升了可鉴别特征提取能力。然后以K近邻算法作为度量学习器实现小样本数据的匹配和分类。最后通过改进中心损失进一步提升特征表示的分辨能力,并通过情景训练从实验室构建的大样本集中学习可迁移知识。实验结果表明,本文方法在每类支持集样本数为5时便可达到94.58%的诊断精度,相对于卷积神经网络提升了63.71%。同时,得益于情景训练方式本文方法有效避免了非平衡样本的问题。     

基于BP神经网络的小型组足球机器人视觉子系统的图像分割技术

为解决小型足球机器人视觉子系统图像分割的实时性和光照适应性问题,将BP神经网络技术应用到图像分割中.在图像分割技术和BP神经网络的理论分析基础上,建立了两者之间的关系,并建立了相应的BP神经网络模型.图像像素离散化并将其H、Cb、Cr分量值作为神经网络的输入,将目标像素点分类类别作为神经网络的输出.通过改进神经网络学习...     

A hybrid deep learning model for robust prediction of the dimensional accuracy in precision milling of thin-walled structural components

The use of artificial intelligence to process sensor data and predict the dimensional accuracy of machined parts is of great interest to the manufacturing community and can facilitate the intelligent production of many key engineering components. In this study, we develop a predictive model of the dimensional accuracy for precision milling of thin-walled structural components. The aim is to classify three typical features of a structural component—squares, slots, and holes—into various categories based on their dimensional errors (i.e., “high precision,” “pass,” and “unqualified”). Two different types of classification schemes have been considered in this study: those that perform feature extraction by using the convolutional neural networks and those based on an explicit feature extraction procedure. The classification accuracy of the popular machine learning methods has been evaluated in comparison with the proposed deep learning model. Based on the experimental data collected during the milling experiments, the proposed model proved to be capable of predicting dimensional accuracy using cutting parameters (i.e., “static features”) and cutting-force data (i.e., “dynamic features”). The average classification accuracy obtained using the proposed deep learning model was 9.55% higher than the best machine learning algorithm considered in this paper. Moreover, the robustness of the hybrid model has been studied by considering the white Gaussian and coherent noises. Hence, the proposed hybrid model provides an efficient way of fusing different sources of process data and can be adopted for prediction of the machining quality in noisy environments.     

基于足底压力传感器与深度学习的生物身份识别

近年来研究表明,足底压力可以反映人体的特征,在生物识别上具有广阔的前景。本研究探讨了足底压力进行身份识别的可行性和方法论。本文使用带有8个压力传感器的鞋垫收集了14名志愿者14 000多个压力数据作为数据集,利用无监督学习探讨了分类的科学性,并讨论了地面状况对于压力数据的影响。采用卷积神经网络(CNN)作为分类器,对分类性能进行了评价,研究了步态分割和多步态周期对提高精度的影响。实验结果表明,使用分割后的数据分类准确率为98.8%,而没有分割的为93.6%,当使用3个和5个步态周期进行分类时,准确率提升到99.4%和99.8%。结果表明,用CNN对分割后的数据并选择多个步态周期进行分类在利用足底压力进行身份识别方面具有良好实用价值。     

Multilevel thresholding for segmentation of medical brain images using real coded genetic algorithm

Medical image analysis is one of the major research areas in the last four decades. Many researchers have contributed quite good algorithms and reported results. In this paper, real coded genetic algorithm with Simulated Binary Crossover (SBX) based multilevel thresholding is used for the segmentation of medical brain images. The T2 weighted Magnetic Resonance Imaging (MRI) brain images are considered for image segmentation. The optimum multilevel thresholding is found by maximizing the entropy. The results are compared with the results of the existing algorithms like Nelder–Mead simplex, PSO, BF and ABF. The statistical performances of the 100 independent runs are reported. The results reveal that the performance of real coded genetic algorithm with SBX crossover based optimal multilevel thresholding for medical image is better and has consistent performance than already reported methods.     

采用深度级联卷积神经网络的三维点云识别与分割

三维目标识别和模型语义分割在自动驾驶、机器人导航、3D打印和智能交通等领域均有着广泛应用。针对PointNet++未能结合三维模型的上下文几何结构信息的问题,提出一种采用深度级联卷积神经网络的三维点云识别与分割方法。首先,通过构建深度动态图卷积神经网络捕捉点云的深层语义几何特征;其次,通过将深度动态图卷积神经网络作为深度级联卷积神经网络的子网络递归地应用于输入点集的嵌套分区,以充分挖掘三维模型的深层细粒度几何特征;最后,针对点集特征学习中的点云采样不均匀问题,构建一种密度自适应层,利用循环神经网络编码每个采样点的多尺度邻域特征以捕捉上下文细粒度几何特征。实验结果表明,本算法在三维目标识别数据集ModelNet40和MoelNet10上的识别准确率分别为91.9%和94.3%,在语义分割数据集ShapeNet Part,S3DIS和vKITTI上的平均交并比分别为85.6%,58.3%和38.6%。该算法能够提高三维点云目标识别和模型语义分割的准确率,且具有较高的鲁棒性。     

Nuclear spatial and spectral features based evolutionary method for meningioma subtypes classification in histopathology

Meningioma subtypes classification is a real‐world multiclass problem from the realm of neuropathology. The major challenge in solving this problem is the inherent complexity due to high intra‐class variability and low inter‐class variation in tissue samples. The development of computational methods to assist pathologists in characterization of these tissue samples would have great diagnostic and prognostic value. In this article, we proposed an optimized evolutionary framework for the classification of benign meningioma into four subtypes. This framework investigates the imperative role of RGB color channels for discrimination of tumor subtypes and compute structural, statistical and spectral phenotypes. An evolutionary technique, Genetic Algorithm, in combination with Support Vector Machine is applied to tune classifier parameters and to select the best possible combination of extracted phenotypes that improved the classification accuracy (94.88%) on meningioma histology dataset, provided by the Institute of Neuropathology, Bielefeld. These statistics show that computational framework can robustly discriminate four subtypes of benign meningioma and may aid pathologists in the diagnosis and classification of these lesions.     

基于 CNN-LSTM-AM 动态集成模型的电站风机状态预测方法

针对电站负荷变化时风机状态预测模型精度降低的问题,提出一种基于卷积神经网络(CNN) 、长短时记忆( LSTM) 网络与注意力机制(AM)的动态集成状态预测方法。 首先,利用 CNN 将样本数据划分为边界有重叠的不同类别,实现风机 运行状态的软分类;其次,在传统的 LSTM 网络的中引入 AM 层,构造不同工况下的 LSTM-AM 子模型,并将 CNN 输出的软分 类标签作为初始权值,使用遗传算法对权值偏置进行搜索寻优;最后,对各个子模型的输出值加权求和,得到风机不同运行 状态下的集成预测值。 实验结果表明,相较各个 LSTM-AM 子模型和单一 LSTM-AM 模型,本文提出的基于 CNN-LSTM-AM 的 动态集成模型在电站风机变负荷运行时可以将预测结果的均方根误差分别减小 11. 5% 和 22. 3% ,说明此模型具有更好的鲁 棒性和适用性。     

基于多尺度特征融合和抓取质量评估的抓取生成方法

在非结构化场景中，物体的6-Ddf抓取是智能服务机器人领域的一项极具挑战性的任务。在该场景中，机器人需要应对不同大小和形状的物体以及环境噪声等因素的干扰，因此难以生成准确的抓取姿态。针对此问题本文提出一种基于多尺度特征融合和抓取质量评估的6-Dof抓取姿态生成方法。首先，提出了自适应半径查询法，解决真实环境中点云采样不均匀导致的关键点查询异常的问题；其次设计了一种将多尺度特征和抓取质量评估融合的抓取生成网络，可以生成丰富的6-Dof抓取域：最后，定义了一种抓取质量评估方法，包含力闭合分数、接触面平整度、棱边分析和质心分数，并将这些标准应用在标准数据集上生成新的抓取置信分数标签，同时将这些标准融人抓取生成网络中。实验结果表明所述的方法与当前较为先进的方法FGC-GraspNet相比平均精度提升了5.9%，单物体抓取成功率提升了5.8%，多物体场景的抓取成功率提升了1.1%。综上所述，本文所提出的方法具备有效性和可行性，在单物体场景和多物体场景中具有较好的适应性。     

基于SSVEP_SSA融合的混合脑机接口研究

针对目前基于体感选择性注意范式的脑机接口控制指令数少,信息传输率低等缺点,提出了一种全新的多模态混合脑机接口系统。该系统融合稳态视觉刺激(SSVEP)和体感选择性注意范式(SSA),在外部视觉和体感刺激的作用下,诱发大脑产生稳态视觉电位和事件相关去同步现象。同时,为了解决传统脑电信息特征提取中需要大量先验知识等问题,引入深度学习算法对混合脑机接口信息进行意图解码,该方法将多通道的时域信息转换成具有时-频-空域三维特征的二维特征图。对8名受试者的离线实验显示,平均识别准确率达到81. 35%,确认了所提出的基于SSVEP_SSA融合的多模态混合脑机接口是可行的,实现了脑机接口(BCI)系统的指令集扩展和高精度解码。     

基于历史误差模型和局域临域校正的预测融合算法

针对预测模型的精度随机性较大,一些数据存在残缺和不确定性,需要融合多模型的组合体来减少预测误差的问题,提出了使用改进模糊RBF神经网络(FRNN)的误差模型的“局域临域校正”的组合体预测融合方法:通过预处理的FRNN误差模型来表征每个预测模型的实时性能,在预测时,利用误差模型,使用基于权值和偏好的融合作为预测模型组合体的聚合方法,将多个预测模型在线融合并给出最终预测结果.该方法将较大的权重赋予局部性能较优的预测模型,同时消除预测偏好的影响,并在预测前预处理所需误差模型.应用结果表明,构建的预测模型其预测精度相比单一预测模型有明显优势,且具有良好的在线预测实时性,具有推广和应用的价值.     

基于深度学习和迁移学习的液压泵健康评估方法

对液压泵建立健康评估模型需要大量训练数据,然而由于其工作条件随时间的地点变化,使得获取特定条件下的数据比较困难。为了在目标数据不足的条件下对液压泵建立健康评估模型,提出了一种深度学习和迁移学习的液压泵健康评估方法。首先,通过卷积神经网络的方法对已有大量历史条件下液压泵振动的频域信号建立预测模型,再用迁移学习的思想在少量目标液压泵数据上对深度学习模型进行微调。实验结果表明,该方法可以有效地提高预测准确率。     

Entropy‐assisted image segmentation for nano‐ and micro‐sized networks

Image analysis is an important tool for characterizing nano/micro network structures. To understand the connection, organization and proper alignment of network structures, the knowledge of the segments that represent the materials inside the image is very necessary. Image segmentation is generally carried out using statistical methods. In this study, we developed a simple and reliable masking method that improves the performance of the indicator kriging method by using entropy. This method selectively chooses important pixels in an image (optical or electron microscopy image) depending on the degree of information required to assist the thresholding step. Reasonable threshold values can be obtained by selectively choosing important pixels in a complex network image composed of extremely large numbers of thin and narrow objects. Thus, the overall image segmentation can be improved as the number of disconnected objects in the network is minimized. Moreover, we also proposed a new method for analyzing high‐pixel resolution images on a large scale and optimized the time‐consuming steps such as covariance estimation of low‐pixel resolution image, which is rescaled by performing the affine transformation on high‐pixel resolution images. Herein, image segmentation is executed in the original high‐pixel resolution image. This entropy‐based masking method of low‐pixel resolution significantly decreases the analysis time without sacrificing accuracy.     

Construction of saliency map and hybrid set of features for efficient segmentation and classification of skin lesion

Skin cancer is being a most deadly type of cancers which have grown extensively worldwide from the last decade. For an accurate detection and classification of melanoma, several measures should be considered which include, contrast stretching, irregularity measurement, selection of most optimal features, and so forth. A poor contrast of lesion affects the segmentation accuracy and also increases classification error. To overcome this problem, an efficient model for accurate border detection and classification is presented. The proposed model improves the segmentation accuracy in its preprocessing phase, utilizing contrast enhancement of lesion area compared to the background. The enhanced 2D blue channel is selected for the construction of saliency map, at the end of which threshold function produces the binary image. In addition, particle swarm optimization (PSO) based segmentation is also utilized for accurate border detection and refinement. Few selected features including shape, texture, local, and global are also extracted which are later selected based on genetic algorithm with an advantage of identifying the fittest chromosome. Finally, optimized features are later fed into the support vector machine (SVM) for classification. Comprehensive experiments have been carried out on three datasets named as PH2, ISBI2016, and ISIC (i.e., ISIC MSK‐1, ISIC MSK‐2, and ISIC UDA). The improved accuracy of 97.9, 99.1, 98.4, and 93.8%, respectively obtained for each dataset. The SVM outperforms on the selected dataset in terms of sensitivity, precision rate, accuracy, and FNR. Furthermore, the selection method outperforms and successfully removed the redundant features.     

一种用于故障分类与预测的多任务特征共享神经网络

智能故障诊断与预测技术在工业实际中得到了广泛地应用,但仍存在以下局限性:1)将不同退化程度的同类型故障作为多种不同的故障模式进行分类识别,脱离了工程应用的实际; 2)基于特定数据训练的诊断模型工况泛化能力差。针对上述问题,提出一种多任务特征共享神经网络,并将其应用于轴承的智能故障诊断与预测。首先,利用卷积神经网络(CNN)构建自适应特征提取器,从原始振动信号中提取深层次特征;其次,同时建立分类与预测的多任务特征共享诊断模型,实现故障类型分类以及故障尺寸预测。最后,通过凯斯西储大学轴承数据集验证了所提方法。试验结果表明:所提方法不但能同时实现对故障类型的分类以及故障尺寸的预测,而且具有较强的工况泛化能力。