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.

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.

基于加权co-occurrence矩阵的聚类集成算法

聚类集成是数据挖掘研究的一个热点。它是利用同一数据集的多个聚类划分集成在一起,以提高聚类分析的性能。当前相关研究大多没有考虑进行集成的聚类成员的质量,因此较差的成员会对集成结果产生不良影响。文中提出了一种基于加权co-occurrence矩阵的聚类集成算法(WCSCE)。该方法首先计算出聚类成员基于属性值的co-occurrence矩阵,然后对聚类成员的质量进行简单评价并赋予权重,生成加权co-occurrence矩阵,进而产生集成结果。最后通过实验验证了该算法的有效性,并提高了聚类质量。

Long noncoding RNAs HAND2-AS1 ultrasound microbubbles suppress hepatocellular carcinoma progression by regulating the miR-873-5p/tissue inhibitor of matrix metalloproteinase-2 axis

BACKGROUND Increasing data indicated that long noncoding RNAs(lncRNAs)were directly or indirectly involved in the occurrence and development of tumors,including hepatocellular carcinoma(HCC).Recent studies had found that the expression of lncRNA HAND2-AS1 was downregulated in HCC tissues,but its role in HCC progression is unclear.Ultrasound targeted microbubble destruction mediated gene transfection is a new method to overexpress genes.AIM To study the role of ultrasound microbubbles(UTMBs)mediated HAND2-AS1 in the progression of HCC,in order to provide a new reference for the treatment of HCC.METHODS In vitro,we transfected HAND2-AS1 siRNA into HepG2 cells by UTMBs,and detected cell proliferation,apoptosis,invasion and epithelial-mesenchymal transition(EMT)by cell counting kit-8 assay,flow cytometry,Transwell invasion assay and Western blotting,respectively.In addition,we transfected miR-837-5p mimic into UTMBs treated cells and observed the changes of cell behavior.Next,the UTMBs treated HepG2 cells were transfected together with miR-837-5p mimic and tissue inhibitor of matrix metalloproteinase-2(TIMP2)overexpression vector,and we detected cell proliferation,apoptosis,invasion and EMT.In vivo,we established a mouse model of subcutaneous transplantation of HepG2 cells and observed the effect of HAND2-AS1 silencing on tumor formation ability.RESULTS We found that UTMBs carrying HAND2-AS1 restricted cell proliferation,invasion,and EMT,encouraged apoptosis,and HAND2-AS1 silencing eliminated the effect of UTMBs.Additionally,miR-873-5p targets the gene HAND2-AS1,which also targets the 3’UTR of TIMP2.And miR-873-5p mimic counteracted the impact of HAND2-AS1.Further,miR-873-5p mimic solely or in combination with pcDNA-TIMP2 had been transformed into HepG2 cells exposed to UTMBs.We discovered that TIMP2 reversed the effect of miR-873-5p mimic caused by the blocked signalling cascade for matrix metalloproteinase(MMP)2/MMP9.In vivo results showed that HAND2-AS1 silencing significantly inhibited tumor formation in mice.CONCLUSION LncRNA HAND2-AS1 promotes TIMP2 expression by targeting miR-873-5p to inhibit HepG2 cell growth and delay HCC progression.

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.

Distribution and occurrence of trace elements in the No.14 coal from the Huolinhe mine

Optical microscopy, and scanning electron microscopy in conjunction with energy dispersed X-ray spectrometry （SEM-EDX）, have been used to study the minerals and the concentrations of 12 trace elements in the No.14 coal from the Huolinhe mine, Inner Mongolia China. The distribution, affinity and removability of the trace elements were studied by float-sink experiments and petrological methods. A high mineral content, dominated by clay minerals, was found in the No.14 coal from the Huolinhe mine. The concentrations of As, Sb and Hg are relatively high compared to the average values for Chinese coals. As, Cr, Hg, Li, Mn, Pb are mainly associated with the minerals while Cd, Co, Ni, Sb, and Se are evenly distributed between the minerals and the organic matter. Be and Ba are mainly distributed in the minerals with a minor proportion in the organic matter. Most elements have a low organic affinity, although Sb, Se, Co, Cd, Ni are closely integrated with the organic matter. High theoretical removabilities are indicated for most trace elements. So it may be possible to lower the concentrations of trace elements during coal preparation.

Altered profiles of nuclear matrix proteins during the differentiation of human gastric mucous adenocarcinoma MGc80-3 cells

AIM： To find and identify specific nuclear matrix proteins associated with proliferation and differentiation of carcinoma cells, which will be potential markers for cancer diagnosis and targets in cancer therapy. METHODS： Nuclear matrix proteins were selectively extracted from MGcS0-3 cells treated with or without hexamethylamine bisacetamide （HMBA）, and subjected to 2-D gel electrophoresis. The resulted protein patterns were analyzed by Melanie software. Spots of nuclear matrix proteins differentially expressed were excised and subjected to in situ digestion with trypsin. Peptide masses were obtained by matrix-assisted laser-desorption/ ionization time of flight mass spectrometry （MALDI-TOFMS） analysis and submitted for database searching using Mascot tool. RESULTS： The MGc80-3 cells were induced into differentiation by HMBA. There were 22 protein spots which changed remarkably in the nuclear matrix, from differentiation of MGcS0-3 cells compared to control. Eleven of which were identified. Seven proteinsactin, prohibitin, porin 31HL, heterogeneous nuclear dbonucleoprotein A2/B1, vimentin, ATP synthase, and heat shock protein 60 were downregulated, whereas three proteins - heat shock protein gp96, heat shock protein 90-beta, and valosin-containing protein were upregulated, and the oxygen-regulated protein was only found in the differentiated MGc80-3 cells. CONCLUSION： The induced differentiation of carcinoma cells is accompanied by the changes of nuclear matrix proteins. Further characterization of those proteins will show the mechanism of cellular proliferation and differentiation, as well as cancer differentiation.

Matrix Solid-phase Dispersion Extraction of Alkaloids from the Roots of Aconitum kusnezoffii Reichb

Matrix solid-phase dispersion（MSPD） was developed for the extraction of four alkaloids, including aconitine, mesaconitine, hypaconitine and deoxyaconitine, from the roots ofAconitum kusnezoffii Reichb. The determination of the analyte was carried out by high performance liquid chromatography with UV detection. The alkaline alumina was used as sorbent. The mixture of acetonitrile and water was used as elution solvent. Several extraction parameters, such as type of sorbent, the ratio of sample to solid support material, type of the elution solvent and the volume of the elution solvent were tested. Mean recoveries ranged from 93.16% to 102.73%, with relative standard deviations from 0.27% to 4.17%. With the extraction efficiency and time expenditure taken into account, MSPD extraction should be a comparatively good method.

Nuclear lamina-like filaments and nuclear matrix in Allium cepa as revealed by scanning electron microscopy

In this study, freeze - fractured specimens of Allium cepa root tip meristems were examined under the scanning electron microscope (SEM). This technique permitted the visualization of the outer membrane of the nuclear envelope with nuclear pore complexes and polyribosomes. Some of the cell nuclei prepared with this procedure had fissures of various widths on their nuclear envelopes through which the nuclear lamina-like filaments (LLF) underneath the nucleoplasmic side of the envelopes were clearly visible. The diameters of these filaments varied between 25 and 125 nm. Many of the LLFs showed granular thickenings at places, and were attached to the inner surface of nuclear envelope in some regions. Similar LLFs were also seen at the peripheries of the freeze -fractured faces of nuclei. Meanwhile,the spatial relation between the nuclear matrix filaments (NMF) and other nuclear structures (nucleoli, chromatin and peripheral lamina - like filaments) was revealed in these fractured preparations. In addition, the methods and techniques in studying the nuclear lamina morphology and the roles played by NMFs in activities of various nuclear structures were discussed in brief.

On 2 - 3 Matrix Chevalley Eilenberg Cohomology

The main objective of this paper is to provide the tool rather than the classical adjoint representation of Lie algebra;which is essential in the conception of the Chevalley Eilenberg Cohomology. We introduce the notion of representation induced by a 2 - 3 matrix. We construct the corresponding Chevalley Eilenberg differential and we compute all its cohomological groups.

基于Mask R-CNN的试管-支架系统Data Matrix码识别方法

在试管-支架自动化系统的输入图像中,Data Matrix(DM)码呈现为多个小目标,图像存在成像模糊、边缘干扰严重等问题,使得传统方法难以达到良好的识别效果。为此,提出一种基于深度学习的Data Matrix码识别方法DeepDMCode,以Mask R-CNN模型为基础,通过内容差异化数据合成和同步自动化标注,实现训练数据的增强,提升模型的学习能力。在模型分割结果的基础上,提出一种旋转校正方法,确保可用标准解码库实现DM码的解码。以分辨率为1600×1200、支架容量为96的数据实验表明,由于该方法在前期码定位阶段最大程度地还原码边界信息,准确度可达0.92(mIoU),完成单张图像中所有DM识别的平均速度为5.2 s,优于YOLO、SegNet、CenterNet等主流工业基准算法。

基于灰度-梯度共生矩阵模型的最大熵阈值处理算法

阈值法是图像分割的一种重要方法 ,在图像处理与识别中广为应用 .本文提出了基于灰度 -梯度共生矩阵模型和最大熵原理的灰度图像的自动阈值化技术 ,该方法不仅利用了图像的灰度信息 ,而且也利用了图像的梯度信息 .该方法通过计算基于灰度 -梯度共生矩阵的二维熵并使边缘区域的熵最大来选择阈值向量 .

基于灰度-梯度共生矩阵和模糊核聚类的振动图形识别方法

以往复机械振动参数图形为对象,提出了基于灰度-梯度共生矩阵和模糊核聚类的振动图形识别方法。利用灰度-梯度共生矩阵直接提取振动参数图形中的特征信息,将得到的纹理特征参量作为样本输入空间,通过Mercer核把输入样本映射到高斯特征空间后,在高维特征空间中进行聚类,从而实现往复机械故障智能诊断。实验结果表明,该方法可以获得较高的诊断精度,具有一定的可行性和有效性。

强直性脊柱炎患者血清骨桥蛋白及基质金属蛋白酶-3检测临床意义研究

目的探讨强直性脊柱炎(AS)患者血清骨桥蛋白(OPN)及基质金属蛋白酶-3(MMP-3)水平的变化及其临床意义。方法采用酶联免疫吸附法检测并比较50例AS患者及20例健康者血清OPN及MMP-3水平,并与患者血细胞沉降率(ESR)、C反应蛋白(CRP)水平进行相关分析。结果 AS患者血清OPN水平明显高于健康者(P<0.05),且与患者ESR及CRP水平呈正相关(P<0.05);AS患者血清MMP-3水平明显高于健康者(P<0.05),且与患者ESR、CRP水平呈正相关(P<0.05);AS患者血清OPN水平与MMP-3水平呈正相关(P<0.05)。结论 OPN及MMP-3参与AS发病,其血清水平检测可用于患者病情活动判断及疗效评价。

融合共现推理的Faster R-CNN输电线路金具检测

为促进目标检测模型与电力领域业务知识有机融合,缓解金具样本间长尾分布问题,有效提升输电线路金具的自动化检测效果,提出了融合共现推理的Faster R-CNN(faster region-based convolutional neural network)输电线路金具检测模型。首先,深入研究输电线路金具结构化组合规则,通过数据驱动的方式以条件概率对金具目标间的共现连接关系进行有效表达;然后,结合图学习方法,利用学习并映射的共现概率关联作为共现图邻接矩阵,使用基线模型(Faster R-CNN)提取的特征向量作为图推理输入特征,以自学习的变换矩阵作为共现图传播权重,完成图信息传播并实现有效的共现推理模型嵌入。实验证明,融合共现推理模块的Faster R-CNN模型较原始模型提升了6.56%的准确率,对于长尾分布样本中数量较少的金具性能提升尤其显著。

基于K-L变换和灰度共生矩阵下的桂西—滇东南地质构造遥感信息提取方法研究

桂西—滇东南是植被高覆盖区域,地质遥感信息相对于地表覆盖(植被)信息而言相对较弱。有效消除地表覆盖信息、突出地质遥感信息是遥感地质专题信息提取的瓶颈之一。本文运用K-L变换提取代表地质构造信息的成分波段,有效滤去地表覆盖信息;利用灰度共生矩阵、定向检测和环形滤波检测相结合的方法,并通过阈值设置,提取地质构造的遥感专题信息,从而达到实现计算机自动提取地质构造信息的目的。经过野外实地验证,结果表明:地质构造信息提取精度较高,特别是一些小的地质构造信息也能有效提取;但是环形地质构造信息没有有效提取,地表覆盖类型对地质构造信息的提取仍有较小影响。

基于灰度-梯度共生矩阵的串联故障电弧特征

为了获取非线性负载回路中串联故障电弧的特征,利用自行研制的故障电弧实验系统开展了变频器和工控机负载条件下的串联故障电弧实验。提出一种基于灰度-梯度共生矩阵的特征提取方法。采用前向差分法对电流信号进行预处理,并利用小波包技术对差分信号进行分解、重构,将重构信号按频率高低依次放入二维数组;二维数组中同时刻各频段的能量归一化后,将其转换为0~255的灰度值图像,对灰度值图像进行Wiener滤波并采用Laplace算子进行锐化和加强处理,取频率高于1 562.5 Hz的图像求解灰度-梯度共生矩阵。通过计算共生矩阵的15种特征量,筛选得到工控机和变频器负载条件下串联故障电弧的特征向量。将获取的故障电弧特征向量输入到支持向量机进行故障电弧识别测试,验证了上述故障电弧特征提取方法的有效性。

基于灰度-梯度共生矩阵的钢轨表面缺陷检测方法

根据钢轨表面缺陷的灰度和梯度特征,提出了钢轨缺陷自动检测方法:基于灰度-梯度共生矩阵为模型提取钢轨缺陷的内边缘,其中以最大熵方法自动求取灰度-梯度二维阈值向量,利用形态学方法对分割后的二值图像进行后期处理,有效区分正常轨面、缺陷区域、阴影和干扰区域.实验表明:本方法能较好地对钢轨表面缺陷进行检测.

基于灰度-梯度共生矩阵的旋转机械振动时频图形识别方法

针对旋转机械振动图形信息在故障诊断中一直没有得到充分利用,在一定程度上影响诊断技术的推广和利用的问题。研究了基于灰度-梯度共生矩阵的旋转机械故障诊断方法,该方法利用能反映图形像素点灰度与梯度的分布规律的灰度-梯度共生矩阵,直接提取和挖掘旋转机械振动信号的时频灰度图形中的像素点与其邻域像素点空间关系的特征信息,有效地提取图形中纹理特征信息后,直接利用人工免疫反面选择算法实现旋转机械故障诊断。在600 MW模化汽轮机转子试验台上进行了转子正常、转子不平衡故障、转子不对中故障及轴承松动故障的试验,诊断结果表明可以获得较高的诊断精度,并与基于灰度共生矩阵的诊断方法进行了比较,证明该方法可以提高故障诊断的准确率,验证了该方法的可行性。