Influence of age and osteoporosis on calcaneus trabecular bone structure: a preliminary in vivo MRI study by quantitative texture analysis

Recent developments in high-resolution MR imaging techniques have opened up new perspectives for structural characterization of trabecular bone by non-invasive methods. In this study, 3-D MR imaging was performed on 17 healthy volunteers and 6 osteoporotic patients. Two different MR sequences were used to evaluate the impact on MR acquisition on texture analysis results. Images were analyzed with four automated methods of texture analysis (grey level histogram, cooccurrence, runlength and gradient matrices) enabling quantitative analysis of grey level intensity and distribution within three different regions of interest (ROI). Texture analysis is not very frequently used since the interpretation of the large number of calculated parameters is difficult. We applied multiparametric data analyses such as principal component analysis (CFA) and hierarchical ascending classification (HAC) to determine the relevant parameters to differentiate between three sets of images (healthy young volunteers, healthy postmenopaused and osteoporotic patients). The results suggest that relevant texture information (depending on the ROI localization in the calcaneus) can be extracted from calcaneus MR images to evaluate osteoporosis and age effects on trabecular bone structure if strictly the same acquisition sequences are used for all patients' examination.     

应用高光谱图像光谱和纹理特征的番茄早疫病早期检测研究

提出了应用光谱和纹理特征的高光谱成像技术早期检测番茄叶片早疫病的方法。利用高光谱图像采集系统获取380～1 030 nm范围内71个染病和88个健康番茄叶片的高光谱图像,同时采用主成分分析法(PCA)对高光谱图像进行处理。选取染病和健康叶片感兴趣区域(region of interest, ROI)的光谱反射率值,同时分别从前8个主成分的每幅主成分图像的ROI中提取对比度(Contrast)、 相关性(Correlation)、 熵(Entropy)和同质性(Homogeneity)4个灰度共生矩阵的纹理特征值,再通过PCA和连续投影算法(SPA)结合最小二乘支持向量机(LS-SVM)构建番茄叶片早疫病的早期鉴别模型。建立的6个模型中,采用光谱反射率值的LS-SVM模型对番茄叶片早疫病的识别率最高,达到100%。结果表明,应用高光谱成像技术检测番茄叶片早疫病是可行的。     

Detection of late epilepsy by the texture analysis of MR brain images in the lithium-pilocarpine rat model

The aim of the study was to detect by texture analysis non easily visible anomalies of magnetic resonance (MR) images of piriform and entorhinal cortices relevant to the lithium-pilocarpine (Li-Pilo) model of temporal lobe epilepsy in rats. Status epilepticus was induced by Li-Pilo in twenty male rats 21 day-old. T₂-weighted MR images of their brain, were obtained before injection of Li-Pilo and one day after status epilepticus. An hyperintense signal was found in the piriform and entorhinal cortices of six rats, which developed chronic epilepsy after a latent period of one to three months. Among the 14 other rats which displayed images similar to those obtained before injection, four remained healthy but 10 rats developed late epileptic symptoms, raising the problem of hidden cortical damage which may be too subtle to be detected by classic MRI examination. A numeric treatment of digital images was then undertaken by texture analysis, to derive image information from a purely computational point of view. The combined texture and discriminant analyses based on pixels pattern anomalies, selected 3 texture parameters derived from co-occurrence matrix which characterized structural abnormalities relevant to the hyperintense signal, not only in the modified images of 6 rats but also in images of 10 rats with apparently non modified images. These three texture’s parameters allowed to classify the twenty rats of our experiment as follows: sixteen epileptic rats were effectively classified with cortical lesions, two non epileptic were correctly classified with healthy cortex, but two healthy rats were not correctly classified. This misclassification is discussed on the basis of the time dependence of the onset of seizure in the Li-Pilo model. These promising results suggest to apply this method to MRI examinations for an improvement of the early diagnostic of human epilepsy.     

Texture analysis of magnetic resonance images of rat muscles during atrophy and regeneration

OBJECTIVES: The goals of the current study were (i) to introduce texture analysis on magnetic resonance imaging (MRI-TA) as a noninvasive method of muscle investigation that can discriminate three muscle conditions in rats; these are normal, atrophy and regeneration; and (ii) to show consistency between MRI-TA results and histological results of muscle type 2 fibers' cross-sectional area. METHOD: Twenty-three adult female Wistar rats were randomized into (i) control (C), (ii) immobilized (I) and (iii) recovering (R) groups. For the last two groups, the right hind limb calf muscles were immobilized against the abdomen for 14 days; then, the hind limb was remobilized only for the R group for 40 days. At the end of each experimental period, MRI was performed using 7-T magnet Bruker Avance DRX 300 (Bruker, Wissembourg); T1-weighted MRI acquisition parameters were applied to show predominantly muscle fibers. Rats were sacrificed, and the gastrocnemius muscle (GM) was excised immediately after imaging. (A) Histology: GM type 2 fibers (fast twitch) were selectively stained using the adenosine triphosphatase (ATPase) technique. The mean cross-sectional areas were compared between the three groups. (B) Image analysis: regions of interest (ROIs) were selected on GM MR images where statistical methods of texture analysis were applied followed by linear discriminant analysis (LDA) and classification. RESULTS: Histological analysis showed that the fibers' mean cross-sectional areas on GM transversal sections represented a significant statistical difference between I and C rats (ANOVA, P<.001) as well as between R and I rats (ANOVA, P<.01), but not between C and R rats. Similarly, MRI-TA on GM transversal images detected different texture for each group with the highest discrimination values (Fisher F coefficient) between the C and I groups, as well as between I and R groups. The lowest discrimination values were found between C and R groups. LDA showed three texture classes schematically separated. CONCLUSION: Quantitative results of MRI-TA were statistically consistent with histology. MRI-TA can be considered as a potentially interesting, reproducible and nondestructive method for muscle examination during atrophy and regeneration.     

Texture analysis of multiple sclerosis: a comparative study

The difficulty of using magnetic resonance imaging (MRI) to support early diagnosis of multiple sclerosis (MS) stems from the subtle pathological changes in the central nervous system (CNS). In this study, texture analysis was performed on MR images of MS patients and normal controls and a combined set of texture features were explored in order to better discriminate tissues between MS lesions, normal appearing white matter (NAWM) and normal white matter (NWM). Features were extracted from gradient matrix, run-length (RL) matrix, gray level co-occurrence matrix (GLCM), autoregressive (AR) model and wavelet analysis, and were selected based on greatest difference between different tissue types. The results of the combined set of texture features were compared with our previous results of GLCM-based features alone. The results of this study demonstrated that (1) with the combined set of texture features, classification was perfect (100%) between MS lesions and NAWM (or NWM), less successful (88.89%) among the three tissue types and worst (58.33%) between NAWM and NWM; (2) compared with GLCM-based features, the combined set of texture features were better at discriminating MS lesions and NWM, equally good at discriminating MS lesions and NAWM and at all three tissue types, but less effective in classification between NAWM and NWM. This study suggested that texture analysis with the combined set of texture features may be equally good or more advantageous than the commonly used GLCM-based features alone in discriminating MS lesions and NWM/NAWM and in supporting early diagnosis of MS.     

Automated artifact detection and removal for improved tensor estimation in motion-corrupted DTI data sets using the combination of local binary patterns and 2D partial least squares

Signal variation in diffusion-weighted images (DWIs) is influenced both by thermal noise and by spatially and temporally varying artifacts, such as rigid-body motion and cardiac pulsation. Motion artifacts are particularly prevalent when scanning difficult patient populations, such as human infants. Although some motion during data acquisition can be corrected using image coregistration procedures, frequently individual DWIs are corrupted beyond repair by sudden, large amplitude motion either within or outside of the imaging plane. We propose a novel approach to identify and reject outlier images automatically using local binary patterns (LBP) and 2D partial least square (2D-PLS) to estimate diffusion tensors robustly. This method uses an enhanced LBP algorithm to extract texture features from a local texture feature of the image matrix from the DWI data. Because the images have been transformed to local texture matrices, we are able to extract discriminating information that identifies outliers in the data set by extending a traditional one-dimensional PLS algorithm to a two-dimension operator. The class-membership matrix in this 2D-PLS algorithm is adapted to process samples that are image matrix, and the membership matrix thus represents varying degrees of importance of local information within the images. We also derive the analytic form of the generalized inverse of the class-membership matrix. We show that this method can effectively extract local features from brain images obtained from a large sample of human infants to identify images that are outliers in their textural features, permitting their exclusion from further processing when estimating tensors using the DWIs. This technique is shown to be superior in performance when compared with visual inspection and other common methods to address motion-related artifacts in DWI data. This technique is applicable to correct motion artifact in other magnetic resonance imaging (MRI) techniques (e.g., the bootstrapping estimation) that use univariate or multivariate regression methods to fit MRI data to a pre-specified model.     

MRI texture analysis on texture test objects, normal brain and intracranial tumors

Texture analysis was performed in three different MRI units on T1 and T2-weighted MR images from 10 healthy volunteers and 63 patients with histologically confirmed intracranial tumors. The goal of this study was a multicenter evaluation of the usefulness of this quantitative approach for the characterization of healthy and pathologic human brain tissues (white matter, gray matter, cerebrospinal fluid, tumors and edema). Each selected brain region of interest was characterized with both its mean gray level values and several texture parameters. Multivariate statistical analyses were then applied in order to discriminate each brain tissue type represented by its own set of texture parameters. Texture analysis was previously performed on test objects to evaluate the method dependence on acquisition parameters and consequently the interest of a multicenter evaluation. Even obtained on different sites with their own acquisition routine protocol, MR brain images contain textural features that can reveal discriminant factors for tissue classification and image segmentation. It can also offer additional information in case of undetermined diagnosis or to develop a more accurate tumor grading.     

Prognosticating brain tumor patient survival after laser thermotherapy: Comparison between neuroradiological reading and semi-quantitative analysis of MRI data

Background and purposeGiven increasing interest in laser interstitial thermotherapy (LITT) to treat brain tumor patients, we explored if examining multiple MRI contrasts per brain tumor patient undergoing surgery can impact predictive accuracy of survival post-LITT.Materials and methodsMRI contrasts included fluid-attenuated inversion recovery (FLAIR), T1 pre-gadolinium (T1pre), T1 post-gadolinium (T1Gd), T2, diffusion-weighted imaging (DWI), apparent diffusion coefficient (ADC), susceptibility weighted images (SWI), and magnetization-prepared rapid gradient-echo (MPRAGE). The latter was used for MRI data registration across preoperative to postoperative scans. Two ROIs were identified by thresholding preoperative FLAIR (large ROI) and T1Gd (small ROI) images. For each MRI contrast, a numerical score was assigned based on changing image intensity of both ROIs (vs. a normal ROI) from preoperative to postoperative stages. The fully-quantitative method was based on changing image intensity across scans at different stages without any human intervention, whereas the semi-quantitative method was based on subjective criteria of cumulative trends across scans at different stages. A fully-quantitative/semi-quantitative score per patient was obtained by averaging scores for each MRI contrast. A standard neuroradiological reading score per patient was obtained from radiological interpretation of MRI data. Scores from all 3 methods per patient were compared against patient survival, and re-examined for comorbidity and pathology effects.ResultsPatient survival correlated best with semi-quantitative scores obtained from T1Gd, ADC, and T2 data, and these correlations improved when biopsy and comorbidity were included.ConclusionThese results suggest interfacing neuroradiological readings with semi-quantitative image analysis can improve predictive accuracy of patient survival.     

Pixel-by-pixel analysis of DCE MRI curve patterns and an illustration of its application to the imaging of the musculoskeletal system

Dynamic contrast enhanced (DCE) MRI is a widespread method that has found broad application in the imaging of the musculoskeletal (MSK) system. A common way of analyzing DCE MRI images is to look at the shape of the time-intensity curve (TIC) in pixels selected after drawing an ROI in a highly enhanced area. Although often applied to a number of MSK affections, shape analysis has so far not led to a unanimous correlation between these TIC patterns and pathology. We hypothesize that this might be a result of the subjective ROI approach. To overcome the shortcomings of the ROI approach (sampling error and interuser variability, among others), we created a method for a fast and simple classification of DCE MRI where time-curve enhancement shapes are classified pixel by pixel according to their shape. The result of the analysis is rendered in multislice, 2D color-coded images. With this approach, we show not only that differences on a short distance range of the TIC patterns are significant and cannot be appreciated with a conventional ROI analysis but also that the information that shape maps and conventional standard DCE MRI parameter maps convey are substantially different.     

A 3D densely connected convolution neural network with connection-wise attention mechanism for Alzheimer's disease classification

PurposeAlzheimer's disease (AD) is a progressive and irreversible neurodegenerative disease. In recent years, machine learning methods have been widely used on analysis of neuroimage for quantitative evaluation and computer-aided diagnosis of AD or prediction on the conversion from mild cognitive impairment (MCI) to AD. In this study, we aimed to develop a new deep learning method to detect or predict AD in an efficient way.Materials and methodsWe proposed a densely connected convolution neural network with connection-wise attention mechanism to learn the multi-level features of brain MR images for AD classification. We used the densely connected neural network to extract multi-scale features from pre-processed images, and connection-wise attention mechanism was applied to combine connections among features from different layers to hierarchically transform the MR images into more compact high-level features. Furthermore, we extended the convolution operation to 3D to capture the spatial information of MRI. The features extracted from each 3D convolution layer were integrated with features from all preceding layers with different attention, and were finally used for classification. Our method was evaluated on the baseline MRI of 968 subjects from ADNI database to discriminate (1) AD versus healthy subjects, (2) MCI converters versus healthy subjects, and (3) MCI converters versus non-converters.ResultsThe proposed method achieved 97.35% accuracy for distinguishing AD patients from healthy control, 87.82% for MCI converters against healthy control, and 78.79% for MCI converters against non-converters. Compared with some neural networks and methods reported in recent studies, the classification performance of our proposed algorithm was among the top ranks and improved in discriminating MCI subjects who were in high risks of conversion to AD.ConclusionsDeep learning techniques provide a powerful tool to explore minute but intricate characteristics in MR images which may facilitate early diagnosis and prediction of AD.     

Utilization of the nephrectomized mouse for determining threshold effects of MRI contrast agents

The tissue concentration of an extravascularly distributed MRI contrast agent required to achieve a 20% change in the MRI signal intensity (SI) of skeletal muscle was determined using radiolabeled gadoteridol administered to nephrectomized mice. This minimal change in the quantified SI was reliably detected qualitatively in the MR muscle images. MR images of muscle were acquired following each intravenous injection of six sequential doses of 0.8 micromol of 153Gd-labeled gadoteridol. A 2.0 T imaging spectrometer and a T1-weighted spin-echo pulse sequence were used to acquire the MR images. After imaging, the injected 153Gd in muscle was measured, and the 153Gd assay results were used to determine the gadoteridol concentration in muscle following each injection. The muscle concentrations of gadoteridol were then correlated to the quantified enhanced MR SI of muscle. Using the 20% factor, it was concluded that the amount of gadoteridol necessary to achieve a reliable change in the SI of muscle was 33+/-10 nmol/g-skeletal muscle.     

基于AI+MRI的影像诊断的样本增广与批量标注方法

训练样本是所有领域人工智能（AI）研发的关键因素.目前，基于人工智能+磁共振成像（AI+MRI）的影像诊断存在着训练样本的有效标注数量和类型无法满足研发需求的瓶颈问题.本文利用临床MRI设备对志愿者或阳性病例进行正常或重点病灶区的定量扫描，获取高分辨率各向同性的纵向弛豫时间（T₁）、横向弛豫时间（T₂）、质子密度（Pd）和表观扩散系数（ADC）等物理信息的多维数据矩阵，作为原始数据.开发虚拟MRI技术平台，对原始数据（相当于数字人体样本）进行虚拟扫描，实现不同序列不同参数下的多种类磁共振图像输出.选择感兴趣组织具有最好边界区分度的图像种类，经有经验的影像医生对其进行手动勾画并轨迹跟踪形成三维MASK标注矩阵，作为其他种类图像的图像勾画标注模板，从而实现低成本、高效率的MRI样本增广和批量标注.该平台以临床少量阳性病例作为输入，进行样本增广和标注，极大地减少AI对实际扫描样本的要求，降低了影像医生的精力和时间投入，极大地节省了成本，并输出了数量足够的磁共振图像，为基于AI+MRI的影像诊断研发提供低成本的训练数据解决方案.     

Skeletal muscle lymphoma: MRI evaluation.

The magnetic resonance imaging (MRI) features of two cases of malignant lymphoproliferative disease involving skeletal muscle are presented. In both cases involved muscles were quantitatively and subjectively hypointense to fat on T1-weighted spin echo images, hypointense or isointense on T2-weighted spin echo images, and hyperintense on short tau inversion recovery (STIR) images. The findings suggest that lymphoproliferative disease should be considered as an etiology of a skeletal muscle lesion that is hypointense or isointense to fat on T2-weighted spin echo magnetic resonance images.     

高光谱技术融合图像信息的牛肉品种识别方法研究

高光谱图像包含了大量的光谱信息和图像信息,采用高光谱成像技术对牛肉品种进行识别。获取可见-近红外(400~1000 nm)光谱范围内的安格斯牛、利木赞牛、秦川牛、西门塔尔牛、荷斯坦奶牛五个品种共252个牛肉样本的高光谱图像。在ENVI软件中对高光谱图像进行阈值分割并构建掩膜图像,获取样本的感兴趣区域(ROI),并结合伪彩色图对牛肉样本的反射率指数进行可视化表达;采用Kennard-Stone(KS)法对样本集进行划分以提高模型的预测性能;对原始光谱采用卷积平滑(SG)、区域归一化(Area normalize)、基线校正(Baseline)、一阶导数(FD)、标准正态变量变换(SNV)及多元散射校正(MSC)等6种方法进行预处理;采用竞争性自适应重加权算法(CARS)提取特征波长。然后利用颜色矩对不同牛肉样本的颜色特征进行提取;对原始光谱图像进行主成分分析,结合灰度共生矩阵(GLCM)算法,提取主要纹理特征。最后结合偏最小二乘判别(PLS-DA)算法建立牛肉样本基于特征波长、颜色特征以及纹理特征的识别模型。KS法将牛肉样本划分为校正集190个,预测集62个;将未经预处理的光谱数据与经过6种不用预处理的光谱数据进行建模分析,结果发现经FD法处理后的光谱数据所建模型的识别率最高;结合CARS法对经FD法预处理后的光谱数据进行特征波长提取,共提取出22个波长;利用颜色矩和GLCM算法分别提取出每个牛肉样本的9个颜色特征、48个纹理特征。将特征波长数据与颜色、纹理特征信息进行融合建模,结果表明,基于特征光谱+纹理特征的模型识别效果最佳,其校正集与预测集识别率分别为98.42%和93.55%,均高于特征光谱数据模型识别率,说明融合纹理特征后使样本分类信息的表达更加全面;融合颜色特征后模型的校正集识别率均有所增加,但预测集识别率稍逊,颜色特征虽携带了部分有效信息,但这些信息与牛肉样本的相关性不大。因此,寻找与牛肉样本相关性更大的颜色特征是提高模型识别率的重要途径之一。该研究结果为牛肉品种的快速无损识别提供了一定的参考。     

Wavelet-based characterization of vertebral trabecular bone structure from magnetic resonance images at 3 T compared with micro-computed tomographic measurements

Trabecular bone structure and bone density contribute to the strength of bone and are important in the study of osteoporosis. Wavelets are a powerful tool in characterizing and quantifying texture in an image. The purpose of this study was to validate wavelets as a tool in computing trabecular bone thickness directly from gray-level images. To this end, eight cylindrical cores of vertebral trabecular bone were imaged using 3-T magnetic resonance imaging (MRI) and micro-computed tomography (microCT). Thickness measurements of the trabecular bone from the wavelet-based analysis were compared with standard 2D structural parameters analogous to bone histomorphometry (MR images) and direct 3D distance transformation methods (microCT images). Additionally, bone volume fraction was determined using each method. The average difference in trabecular thickness between the wavelet and standard methods was less than the size of 1 pixel size for both MRI and microCT analysis. A correlation (R) of .94 for microCT measurements and that of .52 for MRI were found for the bone volume fraction. Based on these results, we conclude that wavelet-based methods deliver results comparable with those from established MR histomorphometric measurements. Because the wavelet transform is more robust with respect to image noise and operates directly on gray-level images, it could be a powerful tool for computing structural bone parameters from MR images acquired using high resolution and thus limited signal scenarios.     

利用小波变换和神经网络对罕见病DMD的MRI进行分类识别

杜兴氏肌营养不良(DMD)是一种严重的儿童腿部神经肌肉罕见病。传统的诊断和检测方案一般为有创手段,会带给患儿极大的痛苦。基于受试者的磁共振图像(MRI),采用计算机辅助检测手段探索了有效的无创检测方法。实验分别选用sym4和db4两种小波基函数,对患儿组和健康对照组的MRI进行三种尺度的小波分解,从所得的分解图像中提取12个纹理特征参数,并利用人工神经网络(ANN)算法对图像参数进行分类识别。结果显示:在受试者的两类MRI加权图像(T1和T2)中,T1图像能更好地区分患儿与健康儿童;利用db4函数对图像进行小波分解,其效果略优于sym4函数,且在三种小波分解尺度中,以二层分解最优;利用ANN算法对图像进行分类识别,其灵敏度、特异度和准确率分别高达98.5%、97.3%和97.9%。该处理方法有望为临床提供客观有效的辅助诊断手段,可作为DMD疾病无创检测的尝试探索。     

Evaluation of discrete orthogonal versus polar Stockwell Transform for local multi-resolution texture analysis using brain MRI of multiple sclerosis patients

The Stockwell Transform has the potential to perform multi-resolution texture analysis in magnetic resonance imaging (MRI). However, it is computationally intensive and memory demanding. The polar Stockwell Transform (PST) is rotation-invariant and relatively memory efficient, but still computationally demanding. The new Discrete Orthogonal Stockwell Transform (DOST) appears to have addressed both the computation and storage challenges; however, its utility in localized texture analysis remains unclear. Our goal was to investigate the theory and texture analysis ability of the DOST versus PST using both synthetic and MR images, and explore the relative importance of the associated texture features using a simple classification example based on clinical brain MRI of six multiple sclerosis patients. MRI texture analysis focused on FLAIR images, and the classification used a machine learning algorithm, random forest, that differentiated regions of interest (ROIs) into 2 classes: white matter lesions, and the contralateral normal-appearing white matter (control). Our results showed that the PST features had a greater ability in detecting subtle changes in image structure than the DOST and polar-index DOST (PDOST). Quantitatively, based on 187 lesion and 187 control ROIs, both the PST and the rotation-invariant radial PST performed better in the classification than the DOST and PDOST, where the latter were no better than guessing (p = 0.65 and 0.98). Further analysis using a hierarchical random forest showed that combining MRI signal intensity with the PST or DOST predictions increased the classification performance, with the accuracy, sensitivity, and specificity all improved to >85% in the tests. Collectively, the DOST is less competitive than the PST in localized image texture analysis. The PST features may help with texture-based lesion classification in MS based on clinical brain MRI scans following further verification.     

Study of normal ocular thermogram using textural parameters

Ocular surface temperature (OST) has been studied with numerous approach and Infrared (IR) thermography has proved to be the best way to capture temperature distribution over some surfaces. It is applied to a number of biomedical applications including studies in the field of ophthalmology. However, the analysis of an ocular thermogram is largely in nascent stage, and is usually achieved by first-order texture analysis. This current study conducted second-order texture analysis on ocular thermal images, mainly by cross co-occurrence matrix together with first-order texture analysis, moments and difference histogram. It was found that, for subjects aged above 35 years old their interocular difference in median, textural contrast, moment 2 and moment 3 (in absolute value) were significantly higher than younger peers. Several significant linear correlations among investigated features were observed. The features extracted from cross co-occurrence matrix may play an important role in the diagnosis of ocular diseases.     

基于荧光透射谱和高光谱图像纹理的茶叶病害预测研究

为了实现对茶叶病害的准确预测,避免病害特征提取过程中对茶叶的二次破坏,利用荧光透射技术对茶叶赤叶病叶片的荧光透射光谱特性展开研究。实验采集了健康茶叶叶片样本45个、赤叶病初期叶片样本60个及赤叶病中期叶片样本60个,并按照2∶1的比例划分成训练集和预测集样本数,通过荧光透射手段利用高光谱仪器采集这些叶片的原始荧光透射光谱。通过对这3组叶片样本平均光谱强度曲线的分析,证实了利用荧光透射光谱信息对这3种病害类型叶片进行分类的可行性。然后使用多项式平滑（savitzky-golay, S-G）方法对原始光谱进行平滑和降噪处理。最后采用竞争性自适应重加权抽样法（competitive adaptive reweighted sampling, CARS）对预处理后的光谱数据进行特征波长的选取。经过50次加权采样后,最终选取出4个特征波长,分别为：463,512,586和613 nm。为了最大化提取样本的病害特征信息,强化分类器输入病害特征值的典型性,使用高光谱反射技术,采集4个特征波长下的高光谱图像,分别使用2种不同的纹理提取算法提取病害叶片图像的纹理信息进行对比分析。首先利用灰度共生矩阵（GLCM）提取4幅图像的纹理信息,分别计算4个方向的灰度共生矩阵(0°,45°,90°及135°),然后计算5个共生矩阵的均值和方差。为了提高鲁棒性,取4幅图像纹理信息的平均值作为该叶片的纹理特征值,最终得到10个特征值。利用LBP（local binary patterns）算法获取特征波长下高光谱图像的纹理信息,并使用Uniform模式对LBP模型进行降维,最终每幅图像得到944个维度的LBP特征值,同样取4幅图像的平均值作为该叶片的LBP纹理特征值。最后通过极限学习机（ELM）分别建立特征光谱联合灰度共生矩阵纹理信息及LBP算子纹理信息的预测模型,由于模型的输入特征值不在一个量纲,首先对输入特征值进行归一化处理,然后再定义模型的输出标签,即健康叶片的预测模型输出为1,赤叶病早期为2,中期为3。实验测得基于CARS-GLCM-ELM模型的预测准确率为81.82%,基于CARS-LBP-ELM模型的预测准确率为85.45%,说明利用荧光透射光谱联合LBP算子纹理信息预测效果更好。由于没有达到预期效果,利用Softplus函数对ELM的隐含层激活函数进行了优化,替换掉原来的Sigmod函数,优化后的模型预测分类正确率达到92.73%,基本达到了预期效果。该研究将病害叶片的荧光光谱信息和对应特征波长下高光谱图像的纹理信息进行了融合,研究结果可为茶叶病害的快速、准确预测提供一定的参考价值。