Recognition of Motor Imagery Electroencephalography Using Independent Component Analysis and Machine Classifiers

Motor imagery electroencephalography (EEG), which embodies cortical potentials during mental simulation of left or right finger lifting tasks, can be used to provide neural input signals to activate a brain computer interface (BCI). The effectiveness of such an EEG-based BCI system relies on two indispensable components: distinguishable patterns of brain signals and accurate classifiers. This work aims to extract two reliable neural features, termed contralateral and ipsilateral rebound maps, by removing artifacts from motor imagery EEG based on independent component analysis (ICA), and to employ four classifiers to investigate the efficacy of rebound maps. Results demonstrate that, with the use of ICA, recognition rates for four classifiers (fisher linear discriminant (FLD), back-propagation neural network (BP-NN), radial-basis function neural network (RBF-NN), and support vector machine (SVM)) improved significantly, from 54%, 54%, 57% and 55% to 70.5%, 75.5%, 76.5% and 77.3%, respectively. In addition, the areas under the receiver operating characteristics (ROC) curve, which assess the quality of classification over a wide range of misclassification costs, also improved from .65, .60, .62, and .64 to .74, .76, .80 and .81, respectively.     

Predictive learning with structured (grouped) data

Many applications of machine learning involve sparse and heterogeneous data. For example, estimation of diagnostic models using patients’ data from clinical studies requires effective integration of genetic, clinical and demographic data. Typically all heterogeneous inputs are properly encoded and mapped onto a single feature vector, used for estimating a classifier. This approach, known as standard inductive learning, is used in most application studies. Recently, several new learning methodologies have emerged. For instance, when training data can be naturally separated into several groups (or structured), we can view model estimation for each group as a separate task, leading to a Multi-Task Learning framework. Similarly, a setting where the training data are structured, but the objective is to estimate a single predictive model (for all groups), leads to the Learning with Structured Data and SVM+ methodology recently proposed by Vapnik [(2006). Empirical inference science afterword of 2006. Springer]. This paper describes a biomedical application of these new data modeling approaches for modeling heterogeneous data using several medical data sets. The characteristics of group variables are analyzed. Our comparisons demonstrate the advantages and limitations of these new approaches, relative to standard inductive SVM classifiers.     

In silico prediction of rhabdomyolysis of compounds by self-organizing map and support vector machine

Rhabdomyolysis is a potentially lethal syndrome resulting in leakage of myocyte intracellular contents into the plasma. Some drugs, such as lipid-lowering drugs and antihistamines, can cause rhabdomyolysis. In this work, a dataset containing 186 chemical compounds causing rhabdomyolysis and 117 drugs not causing rhabdomyolysis was collected. The dataset was split into a training set (containing 230 compounds) and a test set (containing 73 compounds). A Kohonen’s self-organizing map (SOM) and a support vector machine (SVM) were applied to develop classification models to differentiate compounds causing and not causing rhabdomyolysis. Using the SOM method, classification accuracies of 93.3% for the training set and 84.5% for the test set were achieved; using the SVM method, classification accuracies of 95.2% for the training set and 84.9% for the test set were achieved. In addition, the extended connectivity fingerprints (ECFP_4) for all the molecules were calculated and analyzed to find the important features of molecules relating to rhabdomyolysis.     

孔隙度方法在肝癌超声图像纹理特征分类中的性能评估

介绍和评估了描述肝癌超声图像纹理特征的孔隙度方法.用14幅正常肝和14幅原发性肝癌图像为样本,将用5种方法得到的正常肝和原发性肝癌图像的孔隙度值进行正态分布检验,5种方法孔隙度值基本都呈正态分布.只有用立方盒质量和盒柱平均值法得到的正常肝和原发性肝癌的孔隙度平均值通过了差异显著性Student-t检验,用这两种方法得到的正常肝孔隙度值有较小的平均值和标准差,原发性肝癌图像孔隙度值有较大的平均值和标准差.对用5种方法得到的孔隙度进行ROC分析结果表明盒柱平均值法得到了最大的ROC曲线下的面积为0.959 2.用10折交叉验证和不同核函数的SVM进行分类,立方盒质量和盒柱平均值法分别得到了96.428 6％和92.857 1％的分类正确率.实验结果表明所提出的盒柱平均值法具有较强的描述肝癌超声图像纹理特征的能力.     

The potential biomarkers for thromboembolism detected by SELDI-TOF-MS

Introduction

Few studies were concerned about searching for specific biomarkers for thromboembolic (arterial and venous) diseases by the use of Surface-Enhanced Laser Desorption/Ionization Time-of-Flight Mass Spectrometry (SELDI-TOF-MS).

Materials and Methods

We screened for potential biomarkers in 69 plasma samples, including samples from 20 patients with idiopathetic deep vein thrombosis (DVT), 20 patients with acute myocardial infarction (AMI), and 29 healthy controls without a history of thromboembolism. Pretreated plasma samples were analyzed on the Protein Biology System IIc plus SELDI-TOF-MS (Ciphergen Biosystems, Fremont, CA). Proteomic spectra of mass to charge ratio (m/z) were generated by the application of plasma to immobilized metal affinity capture (IMAC-3) ProteinChip arrays activated with copper.

Results

A pattern of three biomarkers (m/z: 2 667, 5 914, and 6 890 Da, respectively) with a total accuracy of 100% was selected based on their collective contribution to the optimal separation between patients with AMI and healthy controls. Another spattern consisting of only one biomarker (m/z: 5 914 Da) could totally discriminate patients with DVT and control subjects. For further analysis between patients with AMI and those with DVT, a pattern of four biomarkers (m/z: 3 418, 5 271, 33 378, and 68 125 Da, respectively) was selected with a total accuracy of 82.5%.

Conclusions

Plasma proteomic profiling with SELDI-TOF-MS and ProteinChip technologies provides high sensitivity and specificity in discriminating patients with thrombosis and healthy subjects. The discovered biomarkers might show great potential for early diagnosis of thromboembolic diseases.     

基于PCA及SVM的运动想象脑电信号识别研究

为了解决脑机接口(BCI)中不同意识任务下运动想象脑电信号的分类问题,提出了一种基于PCA及SVM的识别方法。针对Hilbert-Huang变换和AR模型提取的脑电信号特征,首先采用主成分分析PCA对高维特征向量进行降维处理,然后用支持向量机进行分类。最后将本方法分类结果和Fisher线性分类、概率神经网络分类结果进行比较。实验结果表明,该方法分类正确率较高,复杂度低,具有一定的有效性,可用于脑机接口中。     

一种基于SVM的胶囊内窥镜出血智能识别方法

由于Wireless Capsule Endoscopy(WCE)在消化道中采集到的巨大数量的图像均需要医务人员靠肉眼来排查,给医生带来巨大的负担。该文提供了一种基于支持向量机(Support Vector Machine,SVM)分类器的胶囊内窥镜出血智能识别方法,创立一种新的特征参数,并对SVM参数的选择进行实验优化,最终达到94%的特异度与83%的灵敏度。     

基于NIRS技术和PCA-SVM算法6种树脂及其他类中药的快速鉴别

目的:利用近红外漫反射光谱(NIRS)法,结合主成分分析(PCA)和支持向量机(SVM)联用算法,建立6种树脂及其他类中药安息香(Benzoinum),琥珀(Succinum),没药(Myrrha),乳香(Olibanum),松香(Colophonium),天竺黄(Bambusaen Concretio Silicea)的NIR模式识别模型,用于该6味中药的快速鉴别。方法:收集上述6种中药样品,经性状鉴别和理化分析确定正品药材55批,粉碎成均匀粉末,在4 000~12 000 cm~(-1)光谱区,采集各样品粉末的NIR光谱,选取特征谱段9 000~5 400,5 000~4 000 cm~(-1)为建模谱段,分别采用矢量归一化法(vector normalization,VN),一阶导数法(first derivative,FD),二阶导数法(second derivative,SD)3种不同光谱预处理方法进行预处理并分别进行PCA降维。根据主成分空间散点图,优选最佳预处理方法。利用最佳预处理方法处理后光谱的PCA降维数据,建立SVM模式识别模型,SVM模型参数c和g采用网格搜索法结合五折交叉验证进行寻优。对比不同主成分数所建PCA-SVM模型的预测准确率,确定最佳的主成分数,最终建立6种中药NIR快速鉴别模型。结果:在9 000~5 400,5 000~4 000 cm~(-1)建模谱段,确定最佳光谱预处理方法为SD,SD预处理光谱PCA降维后,确定最佳主成分数为3个,累计贡献率达93.57%。经网格搜索法确定最佳SVM建模参数组为c=65 536,g=512。所建PCA-SVM模型对训练集和验证集样品预测正确率均达100%,模型五折交叉验证准确率亦达100%。结论:所建的6种中药NIR光谱PCA-SVM鉴别模型,预测准确率高,模型预测能力强,结合NIRS技术无损、快速的优点,该模型可用于上述6种中药的无损、快速鉴别。     

一种新型算法在组织光学参数测量系统中的应用

无创测量确定生物组织的光学特性参数在医学诊断和治疗领域中有着广泛的应用前景.目前确定组织参数的方法多建立在单层模型条件下,而实际的许多生物组织均具有分层结构,比如在肌肉、颅骨等.因此在多层模型条件下反演计算组织参数具有更大的实际意义.近年许多研究者针对以上问题提出了各种解决方法,如最小二乘法、神经网络方法等,但这些方法都存在需要时间过长或者误差较大的缺点.本文在组织参数测量领域引入数据挖掘办法--支持向量机(support vector machines,SVM),对双层模型中四个待定组织光学参数的确定进行了研究.结果 表明,利用SVM方法确定组织光学参数具有很好的准确性和实时性.