基于改进U-Net的输电线路绝缘子图像分割方法研究

针对输电线路巡检航拍的绝缘子图像存在背景复杂、对比度不明显、图像质量不能保证等情况造成绝缘子分割精度不高的问题,提出一种基于注意力模型改进U-Net网络的分割方法。首先以VGG16替换主干特征提取网络,增强网络的适用性;同时在下采样过程中引入注意力模型,增强对绝缘子目标的辨识能力,抑制背景、噪声等干扰信息,实现更加精确的分割。实验结果表明:CBAM注意力模型与U-Net网络相结合的方式效果最好,平均重叠度可达96.57%。     

基于改进U-net模型的路面裂缝智能识别

路面裂缝快速检测及响应是道路养护部门的一项重要工作,然而传统的裂缝检测方法耗时且准确度低。因此,本文基于改进后的U-net模型实现对路面裂缝精准地自动识别。结合Canny边缘检测、Otsu 阈值分割算法和人为干预手段研发一款半自动标注软件,用以实现路面裂缝的像素级标注。研究以路面2D激光图像为数据集,并在此基础上通过数据增强进行数据集样本扩充,从而构建模型训练原始样本库;在实验分析阶段,使用交叉熵损失函数判断预测值与真实值的误差大小,并结合Adam 算法优化模型。研究表明改进后的U-net模型在识别精度及泛化能力上均优于原U-net模型及全连接神经网络模型。该研究将为道路养护管理部门的路面病害快速检测提供技术支撑,从而利于快速响应、采取措施保证路面的行车安全。     

一种基于超像素和改进U-net的多模态脑部肿瘤图像分割方法

针对多模态脑部肿瘤图像分割难度较大和对脑部肿瘤边缘区域的分割不足等问题,本文将多模态脑部肿瘤图像分割任务分成两部分解决.第一部分是对脑部肿瘤轮廓区域的分割,先用超像素分割算法对图像进行预处理简化图像的表示形式,再提取每个超像素区域的灰度直方图,通过皮尔逊相关系数计算每个超像素区域的相似度,最后用均值漂移算法对剩余的直方...     

基于随机森林的宫颈鳞癌放化疗疗效预测

对ⅡB~IVA期的宫颈鳞癌患者来说,放化疗治疗后肿瘤区域可能会出现完全缓解和不完全缓解的情况.根据临床经验可知,如果放化疗后肿瘤区域不能完全缓解,那么患者的生存率很低,而且再采取手术治疗或口服靶向药治疗等其他疗法很难有效.因此,在治疗前筛选出对放化疗不敏感的患者,转而探索个性化治疗方案很有必要.针对上述问题,将放化疗疗效预测问题视为图像分类问题,提出一种基于随机森林算法的宫颈鳞癌放化疗疗效预测模型,筛选出对放化疗不敏感的患者.该模型首先利用小波变换和高斯拉普拉斯算子对3D宫颈鳞癌MRI（magnetic resonance imaging）进行预处理;其次,利用U-net分割宫颈鳞癌MR图像中肿瘤区域;再次,结合3D宫颈鳞癌MR图像和相应的肿瘤区域分割结果提取宫颈鳞癌病变区域的纹理及形状特征,并对提取的特征进行筛选,训练随机森林模型.实验数据集由已标记的85位宫颈鳞癌ⅡB~IVA期患者治疗前MR图像序列组成.实验结果表明,基于随机森林算法的疗效预测模型预测宫颈鳞癌放化疗疗效AUC值为0.921,优于目前最先进的预测方法.     

基于改进U-net的肺癌识别方法

目前基于深度学习的肺癌辅助诊断方法存在无法准确定位病灶的缺陷。针对该问题,在现有U-net网络结构的基础上提出一种分两步走的基于改进U-net的肺癌识别方法。利用U-net获得病灶精确位置,通过CNN分类网络对病灶进行诊断,得到原始CT图像的检测结果。实验结果表明,该方法可以对肺部病灶进行较为精确的定位,分割效果的DSC相似度指数超过80%,对肺癌病灶进行分类诊断的准确率达到90.7%。     

Research on the magnetic resonance imaging brain tumor segmentation algorithm based on DO-UNet

With the social and economic development and the improvement of people's living standards, smart medical care is booming, and medical image processing is becoming more and more popular in research, of which brain tumor segmentation is an important branch of medical image processing. However, the manual segmentation method of brain tumors requires a lot of time and effort from the doctor and has a great impact on the treatment of patients. In order to solve this problem, we propose a DO-UNet model for magnetic resonance imaging brain tumor image segmentation based on attention mechanism and multi-scale feature fusion to realize fully automatic segmentation of brain tumors. Firstly, we replace the convolution blocks in the original U-Net model with the residual modules to prevent the gradient disappearing. Secondly, the multi-scale feature fusion is added to the skip connection of U-Net to fuse the low-level features and high-level features more effectively. In addition, in the decoding stage, we add an attention mechanism to increase the weight of effective information and avoid information redundancy. Finally, we replace the traditional convolution in the model with DO-Conv to speed up the network training and improve the segmentation accuracy. In order to evaluate the model, we used the BraTS2018, BraTS2019, and BraTS2020 datasets to train the improved model and validate it online, respectively. Experimental results show that the DO-UNet model can effectively improve the accuracy of brain tumor segmentation and has good segmentation performance.     

融合门控自注意力机制的生成对抗网络视频异常检测

目的 视频异常行为检测是当前智能监控技术的研究热点之一,在社会安防领域具有重要应用。如何通过有效地对视频空间维度信息和时间维度信息建模来提高异常检测的精度仍是目前研究的难点。由于结构优势,生成对抗网络目前广泛应用于视频异常检测任务。针对传统生成对抗网络时空特征利用率低和检测效果差等问题,本文提出一种融合门控自注意力机制的生成对抗网络进行视频异常行为检测。方法 在生成对抗网络的生成网络U-net部分引入门控自注意力机制,逐层对采样过程中的特征图进行权重分配,融合U-net网络和门控自注意力机制的性能优势,抑制输入视频帧中与异常检测任务不相关背景区域的特征表达,突出任务中不同目标对象的相关特征表达,更有效地针对时空维度信息进行建模。采用LiteFlownet网络对视频流中的运动信息进行提取,以保证视频序列之间的连续性。同时,加入强度损失函数、梯度损失函数和运动损失函数加强模型检测的稳定性,以实现对视频异常行为的检测。结果 在CUHK （Chinese University of Hong Kong） Avenue、UCSD （University of California,San Diego） Ped1和UCSD Ped2等视频异常事件数据集上进行实验。在CUHK Avenue数据集中,本文方法的AUC （area under curve）为87.2%,比同类方法高2.3%;在UCSD Ped1和UCSD Ped2数据集中,本文方法的AUC值均高于同类其他方法。同时,设计了4个消融实验并对实验结果进行对比分析,本文方法具有更高的AUC值。结论 实验结果表明,本文方法更适合视频异常检测任务,有效提高了异常行为检测任务模型的稳定性和准确率,且采用视频序列帧间运动信息能够显著提升异常行为检测性能。     

基于U-net模型的全自动鼻咽肿瘤MR图像分割

鼻咽肿瘤生长方向不确定，解剖结构复杂，当前主要依靠医生手动分割，该方法耗时久同时严重依赖于医生的经验。针对这一问题，基于深度学习理论，提出一种基于U-net模型的全自动鼻咽肿瘤MR图像分割算法，利用卷积操作替换原始U-net模型中的最大池化操作以减少特征信息的损失。首先，从所有患者的肿瘤切片中提取大小为128×128的区域作为数据样本；然后，将患者样本分为训练样本集和测试样本集，并对训练样本集进行数据扩充；最后，选择训练样本集中所有数据用于训练网络模型。为了验证所提模型的有效性，选取测试样本集中患者的所有肿瘤切片进行分割，最终平均分割精度可达到：DSC（Dice Similarity Coefficient）为80.05%，PM系数为85.7%，CR系数为71.26%，ASSD（Average Symmetric Surface Distance）指标为1.1568。与基于图像块的卷积神经网络（CNN）相比，所提算法DSC，PM（Prevent Match）、CR（Correspondence Ratio）系数分别提高了9.86个百分点、19.61个百分点、16.02个百分点，ASSD指标下降了0.4364；与全卷积神经网络（FCN）模型及基于最大池化的U-net网络相比，所提算法的DSC、CR系数均取得了最优结果，PM系数较两种对比模型中的最大值低2.55个百分点，ASSD指标较两种对比模型中的最小值略高出0.0046。实验结果表明，所提算法针对鼻咽肿瘤图像可以实现较好的自动化分割效果以辅助医生进行诊断。     

Detection of Precipitation Cloud over the Tibet Based on the Improved U-Net

Aiming at the problem of radar base and ground observation stations on the Tibet is sparsely distributed and cannot achieve large-scale precipitation monitoring. UNet, an advanced machine learning (ML) method, is used to develop a robust and rapid algorithm for precipitating cloud detection based on the new-generation geostationary satellite of FengYun-4A (FY-4A). First, in this algorithm, the real-time multi-band infrared brightness temperature from FY-4A combined with the data of Digital Elevation Model (DEM) has been used as predictor variables for our model. Second, the efficiency of the feature was improved by changing the traditional convolution layer serial connection method of U-Net to residual mapping. Then, in order to solve the problem of the network that would produce semantic differences when directly concentrated with low-level and high-level features, we use dense skip pathways to reuse feature maps of different layers as inputs for concatenate neural networks feature layers from different depths. Finally, according to the characteristics of precipitation clouds, the pooling layer of U-Net was replaced by a convolution operation to realize the detection of small precipitation clouds. It was experimentally concluded that the Pixel Accuracy (PA) and Mean Intersection over Union (MIoU) of the improved U-Net on the test set could reach 0.916 and 0.928, the detection of precipitation clouds over Tibet were well actualized.     

Pediatric brain extraction from T2-weighted MR images using 3D dual frame U-net and human connectome database

Accurate extraction of brain tissues from magnetic resonance (MR) images is important in neuroradiology. However, brain extraction is more difficult for pediatric brains than for adult brains due to several factors including smaller brain sizes and lower tissue contrasts. In this work, we propose a brain extraction technique that utilizes dual frame (DF) 3D U-net deep learning architecture and the human connectome project (HCP) database for multislice 2D pediatric T2-weighted MR images with diseases. To improve segmentation accuracy in small pediatric brains with detailed boundary regions, DF 3D U-net architecture was used. We pretrained networks with the HCP database to compensate for the limited amount of MR images and manual segmentation masks of pediatric patients. For quantitative analysis, we compared the brain extraction results of brain extraction tool, DF, and conventional 3D U-net using the dice similarity coefficient (DSC), intersection of union (IoU), and boundary F1 (BF) scores; each deep learning architecture was evaluated with and without pretraining using the HCP. This study included 10 patients with diseases and all images were acquired using a PROPELLER MR sequence. Pretraining using the HCP database enhanced segmentation performance of the network, and the skip connections in the DF 3D U-net could enhance the contour similarity of segmentation results. Experimental results showed that the proposed method increased the DSC, IoU, and BF scores by 0.8%, 1.6%, and 1.5%, respectively, compared with those of the conventional 3D U-net without pretraining.