Machine learning prediction model for gray-level co-occurrence matrix features of synchronous liver metastasis in colorectal cancer

BACKGROUND Synchronous liver metastasis(SLM)is a significant contributor to morbidity in colorectal cancer(CRC).There are no effective predictive device integration algorithms to predict adverse SLM events during the diagnosis of CRC.AIM To explore the risk factors for SLM in CRC and construct a visual prediction model based on gray-level co-occurrence matrix(GLCM)features collected from magnetic resonance imaging(MRI).METHODS Our study retrospectively enrolled 392 patients with CRC from Yichang Central People’s Hospital from January 2015 to May 2023.Patients were randomly divided into a training and validation group(3:7).The clinical parameters and GLCM features extracted from MRI were included as candidate variables.The prediction model was constructed using a generalized linear regression model,random forest model(RFM),and artificial neural network model.Receiver operating characteristic curves and decision curves were used to evaluate the prediction model.RESULTS Among the 392 patients,48 had SLM(12.24%).We obtained fourteen GLCM imaging data for variable screening of SLM prediction models.Inverse difference,mean sum,sum entropy,sum variance,sum of squares,energy,and difference variance were listed as candidate variables,and the prediction efficiency(area under the curve)of the subsequent RFM in the training set and internal validation set was 0.917[95%confidence interval(95%CI):0.866-0.968]and 0.09(95%CI:0.858-0.960),respectively.CONCLUSION A predictive model combining GLCM image features with machine learning can predict SLM in CRC.This model can assist clinicians in making timely and personalized clinical decisions.

Artificial intelligence on diabetic retinopathy diagnosis: an automatic classification method based on grey level co-occurrence matrix and naive Bayesian model

AIM: To develop an automatic tool on screening diabetic retinopathy(DR) from diabetic patients.METHODS: We extracted textures from eye fundus images of each diabetes subject using grey level co-occurrence matrix method and trained a Bayesian model based on these textures. The receiver operating characteristic(ROC) curve was used to estimate the sensitivity and specificity of the Bayesian model.RESULTS: A total of 1000 eyes fundus images from diabetic patients in which 298 eyes were diagnosed as DR by two ophthalmologists. The Bayesian model was trained using four extracted textures including contrast, entropy, angular second moment and correlation using a training dataset. The Bayesian model achieved a sensitivity of 0.949 and a specificity of 0.928 in the validation dataset. The area under the ROC curve was 0.938, and the 10-fold cross validation method showed that the average accuracy rate is 93.5%.CONCLUSION: Textures extracted by grey level cooccurrence can be useful information for DR diagnosis, and a trained Bayesian model based on these textures can be an effective tool for DR screening among diabetic patients.

Binary Image Steganalysis Based on Distortion Level Co-Occurrence Matrix

In recent years,binary image steganography has developed so rapidly that the research of binary image steganalysis becomes more important for information security.In most state-of-the-art binary image steganographic schemes,they always find out the flippable pixels to minimize the embedding distortions.For this reason,the stego images generated by the previous schemes maintain visual quality and it is hard for steganalyzer to capture the embedding trace in spacial domain.However,the distortion maps can be calculated for cover and stego images and the difference between them is significant.In this paper,a novel binary image steganalytic scheme is proposed,which is based on distortion level co-occurrence matrix.The proposed scheme first generates the corresponding distortion maps for cover and stego images.Then the co-occurrence matrix is constructed on the distortion level maps to represent the features of cover and stego images.Finally,support vector machine,based on the gaussian kernel,is used to classify the features.Compared with the prior steganalytic methods,experimental results demonstrate that the proposed scheme can effectively detect stego images.

3D Gray Level Co-Occurrence Matrix Based Classification of Favor Benign and Borderline Types in Follicular Neoplasm Images

Since the efficiency of treatment of thyroid disorder depends on the risk of malignancy, indeterminate follicular neoplasm (FN) images should be classified. The diagnosis process has been done by visual interpretation of experienced pathologists. However, it is difficult to separate the favor benign from borderline types. Thus, this paper presents a classification approach based on 3D nuclei model to classify favor benign and borderline types of follicular thyroid adenoma (FTA) in cytological specimens. The proposed method utilized 3D gray level co-occurrence matrix (GLCM) and random forest classifier. It was applied to 22 data sets of FN images. Furthermore, the use of 3D GLCM was compared with 2D GLCM to evaluate the classification results. From experimental results, the proposed system achieved 95.45% of the classification. The use of 3D GLCM was better than 2D GLCM according to the accuracy of classification. Consequently, the proposed method probably helps a pathologist as a prescreening tool.

A Combination of Feature Selection and Co-occurrence Matrix Methods for Leukocyte Recognition System

A leukocyte recognition system, as part of a differential blood counter system, is very important in hematology field. In this paper, the propose system aims to automatically classify the white blood cells (leukocytes) on a given microscopic image. The classifications of leukocytes are performed based on the combination of color and texture features of the blood cell images. The developed system classifies the leukocytes in one of the five categories (neutrophils, eosinophils, basophils, lymphocytes, and monocytes). In the preprocessing stage, the system starts with converting the microscopic images from Red Green Blue (RGB) color space to Hue Saturation Value (HSV) color space. Next, the system splits the Hue and Saturation features from the Value feature. For both Hue and Saturation features, the system processes their color information using the Feature Selection method and the Window Cropping method;while the Value feature is processed by its texture information using the Co-occurrence matrix method. The final recognition stage is performed using the Euclidean distance method. The combination of the Feature Selection and Co-occurrence Matrix methods gives the best overall recognition accuracies for classifying leukocyte images.

Material microstructures analyzed by using gray level Co-occurrence matrices

The mechanical properties of materials greatly depend on the microstructure morphology. The quantitative characterization of material microstructures is essential for the performance prediction and hence the material design. At present,the quantitative characterization methods mainly rely on the microstructure characterization of shape, size, distribution,and volume fraction, which related to the mechanical properties. These traditional methods have been applied for several decades and the subjectivity of human factors induces unavoidable errors. In this paper, we try to bypass the traditional operations and identify the relationship between the microstructures and the material properties by the texture of image itself directly. The statistical approach is based on gray level Co-occurrence matrix（GLCM）, allowing an objective and repeatable study on material microstructures. We first present how to identify GLCM with the optimal parameters, and then apply the method on three systems with different microstructures. The results show that GLCM can reveal the interface information and microstructures complexity with less human impact. Naturally, there is a good correlation between GLCM and the mechanical properties.

基于L_(2)灵敏度最小化的数字滤波器状态空间实现稀疏化方法

为解决传统数字滤波器在有限精度实现时因有限字长(Finite Word Length,FWL)效应导致滤波器性能下降的问题,提出一种L_(2)灵敏度最小化的数字滤波器状态空间实现稀疏化方法.推导前向差分算子数字滤波器结构传输函数及其等效状态空间实现,根据可控及可观格莱姆矩阵得到基于相似变换矩阵的L_(2)灵敏度表达式,并进行稀疏化校准,将L_(2)灵敏度最小化问题转换为凸函数求最值问题,求导得到L_(2)灵敏度最小化表达式,代回即得前向差分算子数字滤波器的稀疏化状态空间实现.仿真结果表明,所提方法设计的数字滤波器具有更好的抗FWL效应.

高斯混合模型与文本图卷积网络结合的虚假评论识别算法

针对文本图卷积网络(Text GCN)窗口边权阈值策略不足的问题,为了更精准地挖掘相关的词关联结构、提高预测精度,提出一种高斯混合模型(GMM)与Text GCN结合的虚假评论识别算法F-Text GCN。首先,利用GMM分离噪声边权分布的特性,提高虚假评论在训练数据上相对正常评论数不足的边信号强度;然后,考虑到信源的多样性,综合文档、词汇和评论以及非文本特征构造邻接矩阵;最后,通过Text GCN的谱分解提取邻接矩阵的虚假评论关联结构实施预测。根据国内某大型电商平台采集的126086条实际中文评论数据开展实证研究,实验结果表明,F-Text GCN识别虚假评论的F1值达到82.92%,与预训练表征模型BERT和文本卷积神经网络相比分别提升了10.46%和11.60%,相较于只使用评论文本信源的Text GCN模型F1值提升了2.94%;研究了高仿虚假评论的预测错误率,在支持向量机(SVM)作用后难识别的评论样本上尝试二次识别,F-Text GCN整体预测准确率可达94.71%,相较于Text GCN和SVM,在识别准确率上分别提升了2.91%和14.54%。研究发现,虚假评论的二阶图邻居结构显示出较强的干预消费者决策的词汇,这表明所提算法特别适用于提取用于虚假评论检测的长程词语搭配结构和全局句子特征模式变化的场景。

word2vec-ACV:OOV语境含义的词向量生成模型

针对word2vec模型生成的词向量缺乏语境的多义性以及无法创建集外词(OOV)词向量的问题,引入相似信息与word2vec模型相结合,提出word2vec-ACV模型。该模型首先基于连续词袋(CBOW)和Hierarchical softmax的word2vec模型训练出词向量矩阵即权重矩阵;然后将共现矩阵进行归一化处理得到平均上下文词向量,再将词向量组成平均上下文词向量矩阵;最后将平均上下文词向量矩阵与权重矩阵相乘得到词向量矩阵。为了能同时解决集外词及多义性问题,将平均上下文词向量分为全局平均上下文词向量(global ACV)和局部平均上下文词向量(local ACV)两种,并对两者取权值组成新的平均上下文词向量矩阵,并将word2vec-ACV模型和word2vec模型分别进行类比任务实验和命名实体识别任务实验。实验结果表明,word2vec-ACV模型同时解决了语境多义性以及创建集外词词向量的问题,降低了时间消耗,提升了词向量表达的准确性和对海量词汇的处理能力。

微博平台烟草信息的主题特征及情感倾向分析

目的 了解微博用户发布和公众关注的烟草信息的主题特征,并探究各类烟草信息的传播效果,为推动控烟信息的高效传播及效果提升提供理论参考。方法 使用Python软件爬取微博用户发布的烟草信息,采用UCINET软件进行数据分析和共词矩阵分析。结果 经爬取、筛选后共获得6385条烟草信息。其中,烟草信息的关键词整体呈高异质性、广覆盖特征,其频次以“吸烟”“抽烟”居多;正向情感的烟草信息关键词(68.71%)较负向情感(6.98%)多,其频次以“吸烟”“二手烟”居多。结论 微博平台中烟草信息的网络生态整体较好,烟草相关内容占比较小,但具有较强的诱惑性和隐蔽性。相关部门应持续关注烟草流行词汇,并加强烟草危害和控烟法律的宣传,谨防负向情感的烟草信息传播。

最近对寻址的专利实体关系抽取方法

针对专利领域没有公开数据集的问题,标注一个中文专利实体关系抽取数据集PERD(patent entity relation dataset)。为完成实体关系抽取任务,提出最近对寻址的实体关系抽取模型NPAM(nearest pair addressing entity relationship extraction model),实体位置信息获取方法的改进、注意力机制建模矩阵和实体抽取方法的创新,使该模型在PERD上F1值达到72.74%,相比模型PRGC提升12.64个百分点。实验结果验证了该模型的有效性。

基于自然语言处理的Word2Vec词向量应用

计算机要理解自然语言,首先需要理解词语的语义,要考虑词的同义、近义、词的上下文关系,数字化即转化为词向量,通过计算处理词向量来处理文本。阐述词向量及Word2Vec词模型的特点,Word2Vec是被广泛使用的词向量模型,同时基于《西游记》语料进行应用实现。

基于表调度的Matrix DSP指令调度算法的实现

指令调度是gcc实现指令并行、提高性能的一种优化策略,gcc目前支持的调度算法主要有表调度算法与模调度算法。主要根据Matrix芯片的体系结构特点,对现有的表调度算法进行了改进,实现了Matrix指令调度算法。实验结果表明,改进后的表调度算法能够编译出正确的指令,充分挖掘指令间的并行性,显式标注指令间的并行关系,指令字间的延迟关系符合硬件要求。

相位模糊下基于软判决的卷积码识别方法

QPSK数据流通常具有相位模糊的问题,传统方法常采用卷积码遍历识别码字起始位置的译码算法,这将导致算法计算量大、资源占用高等诸多问题。因此,提出一种基于软判决的卷积码起始位置识别与QPSK相位模糊消除的译码方法,以有效提升译码算法性能。首先,推导了卷积码编码序列与校验矩阵之间的数学表达式,明晰了卷积码码字起始位置与相位模糊的逻辑关系;然后,求解CCSDS标准卷积码的校验矩阵,以方程成立的概率作为判决的度量,该方法适用于各种码率的删除卷积码,具有工程实用性;最后,通过仿真分析验证了所提出方法的有效性。研究表明该方法无需遍历卷积码的码字起始位置和相位模糊,且鲁棒性较强,所需数据量较小,随着数据量的增大和卷积码码率的降低,识别准确率逐渐提高。

在Word文档中插入CAD图形方法的研究

在Word文档中插入CAD图形是工程领域中经常遇到的一个问题,如工厂、企业、院校中,工程技术人员采集数据绘制曲线图像、撰写论文等。现通行的,也是最简单方法是位图抓图法。通过对位图与矢量图两种格式的比较得出用矢量图格式插入CAD图形能更好的保留图形精度和效果。用矢量图格式插入法有内置与外置两种方法。

基于改进的Bag of Visual Words算法的SAR图像目标分类

"视觉词袋"(Bag of Visual Words,BOV)算法是一种有效的基于语义特征表达的物体识别算法。针对传统BOV模型存在的不足,综合利用SAR图像的灰度和纹理特征,提出基于感兴趣目标(Target of Interest,TOI)的"视觉词袋"算法。首先,对训练图像进行TOI选取,用灰度共生矩阵模型提取TOI的纹理特征,再结合灰度特征,组成多维特征向量集,以簇内相似度最高、数据分布密度最大为准则,生成"视觉词袋"。其次,对测试图像,依据已生成的"视觉词袋",采用支持向量机(Support Vector Machine,SVM)分类器,实现SAR图像感兴趣目标的有效分类。实验结果表明,与传统的"视觉词袋"构建算法相比,该算法在分类正确率提高的同时,能够在训练图像较少的情况下达到良好的分类效果。

Non-fragile H∞ Filtering for Discrete-time Systems with Finite Word Length Consideration

过滤有限的词长度(FWL ) 为线性分离时间的系统影响的问题的 nonfragile H 在这份报纸被调查。要设计的过滤器被假定与添加剂获得变化，它在过滤器实现上反映 FWL 效果。结构化的顶点隔板的一个观点被建议处理这个问题并且利用了以一套线性矩阵不平等(LMI ) 为 nonfragile H 过滤器设计开发足够的条件。设计使扩充系统变为 asymptotically 稳定并且保证 H 变细水平不到规定水平。一个数字例子被给说明建议方法的效果。

Direct numerical simulation of matrix diffusion across fracture/matrix interface

Accurate descriptions of matrix diffusion across the fracture/matrix interface are critical to assessing contaminant migration in fractured media. The classical transfer probability method is only applicable for relatively large diffusion coefficients and small fracture spacings, due to an intrinsic assumption of an equilibrium concentration profile in the matrix blocks. Motivated and required by practical applications, we propose a direct numerical simulation （DNS） approach without any empirical assumptions. A three-step Lagrangian algorithm was developed and validated to directly track the particle dynamics across the fracture/matrix interface, where particle＇s diffusive displacement across the discontinuity is controlled by an analytical, one-side reflection probability. Numerical experiments show that the DNS approach is especially efficient for small diffusion coefficients and large fracture spacings, alleviating limitations of the classical modeling approach.

Efficient matrix inversion based on VLIW architecture

Matrix inversion is a critical part in communication, signal processing and electromagnetic system. A flexible and scalable very long instruction word （VLIW） processor with clustered architecture is proposed for matrix inversion. A global register file （RF） is used to connect al the clusters. Two nearby clusters share a local register file. The instruction sets are also designed for the VLIW processor. Experimental results show that the proposed VLIW architecture takes only 45 latency to invert a 4 × 4 matrix when running at 150 MHz. The proposed design is roughly five times faster than the DSP solution in processing speed.

高误码率下基于随机抽取的LDPC码校验矩阵重建

为改善高误码率下LDPC码稀疏校验矩阵重建算法的性能,提出了接收码字个数充足和不充足条件下容错能力较强的校验矩阵开集识别算法。首先,通过多次随机抽取码字的部分比特构建新的码字空间,在较低维度下利用高斯消元法求解对偶向量并还原出校验向量;其次,利用该校验向量,采用“剔除错误码字”或“翻转最低不可靠位”的方法不断提高接收数据内无误码码组的比例进行迭代处理。仿真结果表明,所提算法在不同误码率、不同码长、不同码率、不同码字个数下均优于对比算法。对于IEEE 802.11n协议下的(648,324)LDPC码,当接收码字个数充足时,所提算法在误码率为0.003的条件下,其校验矩阵重建率能达到95%以上;当接收码字个数不足(码字个数为450)时,所提算法在误码率为0.0015的条件下,其校验矩阵重建率能达到90%以上。