Detection of Carotid Artery Disease by Using Learning Vector Quantization Neural Network

Doppler ultrasound has been usually preferred for investigation of the artery conditions in the last two decades, because it is a non-invasive, easy to apply and reliable technique. In this study, a biomedical system based on Learning Vector Quantization Neural Network (LVQ NN) has been developed in order to classify the internal carotid artery Doppler signals obtained from the 191 subjects, 136 of them had suffered from internal carotid artery stenosis and rest of them had been healthy subject. The system is composed of feature extraction and classification parts, basically. In the feature extraction stage, power spectral density (PSD) estimates of internal carotid artery Doppler signals were obtained by using Burg autoregressive (AR) spectrum analysis technique in order to obtain medical information. In the classification stage, LVQ NN was used classify features from Burg AR method. In experiments, LVQ NN based method reached 97.91% classification accuracy with 5 fold Cross Validation (CV) technique. In addition, the classification performance of the LVQ NN was compared with some methods such as Multi Layer Perceptron (MLP) NN, Naive Bayes (NB), K-Nearest Neighbor (KNN), decision tree and Support Vector Machine (SVM) with sensitivity and specificity statistical parameters. The classification results showed that the LVQ NN method is effective for classification of internal carotid artery Doppler signals.     

Breast Tissue Image Classification Based on Semi-supervised Locality Discriminant Projection with Kernels

Breast tissue classification is an important and effective way for computer aided diagnosis of breast cancer. We present Semi-supervised Locality Discriminant Projections with Kernels for breast cancer classification. The contributions of this work lie in: 1) Semi-supervised learning is used into Locality Preserving Projections (LPP) to enhance its performance using side-information together with the unlabelled training samples, while current algorithms only consider the side-information but ignoring the unlabeled training samples. 2) Kernel trick is applied into Semi-supervised LPP to improve its ability in the nonlinear classification. 3) The framework of breast cancer classification with Semi-supervised LPP with kernels is presented. Many experiments are implemented on four breast tissue databases to testify and evaluate the feasibility and affectivity of the proposed scheme.     

基于迁移学习的全连接神经网络舌象分类方法

目的 针对深度学习在舌象分类中训练数据量大、训练设备要求高、训练时间长等问题,提出一种基于迁移学习的全连接神经网络小样本舌象分类方法。方法 应用经ImageNet海量数据集训练后的卷积Inception_v3网络提取舌象点、线等有效特征,再使用全连接神经网络对特征进行训练分类,将深度学习网络学习到的图像知识迁移到舌象识别任务中。利用舌象数据集进行训练、测试。结果 与典型舌象分类方法K最近邻（KNN）算法、支持向量机（SVM）算法和卷积神经网络（CNN）深度学习方法相比,本实验使用的两种方法（Inception_v3+2NN和Inception_v3+3NN）具有较高的舌象分类识别率,准确率分别达90.30%和93.98%,且样本训练时间明显缩短。结论 与KNN算法、SVM算法和CNN深度学习方法相比,基于迁移学习的全连接神经网络舌象分类方法可有效提高舌象分类的准确率、缩短网络的训练时间。     

Mammographic Image Based Breast Tissue Classification with Kernel Self-optimized Fisher Discriminant for Breast Cancer Diagnosis

Breast tissue classification is an important and effective way for computer aided diagnosis of breast cancer with digital mammogram. Current methods endure two problems, firstly pectoral muscle influences the classification performance owing to its texture similar to parenchyma, and secondly classification algorithms fail to deal with the nonlinear problem from the digital mammogram. For these problems, we propose a novel framework of breast tissue classification based on kernel self-optimized discriminant analysis combined with the artifacts and pectoral muscle removal with multi-level segmentation based Connected Component Labeling analysis. Experiments on mini-MIAS database are implemented to testify and evaluate the performance of proposed algorithm.     

A Novel Enhanced Gray Scale Adaptive Method for Prediction of Breast Cancer

Breast cancer is the important problem across the globe in which, most of the women are suffering without knowing the causes and effects of the cancer cells. Mammographic is the most powerful tool for the diagnosis of the Breast cancer. The analysis of this mammogram images proves to be more vital in terms of diagnosis but the accuracy level still needs improvisation. Several intelligent techniques are suggested for the detection of Microcalcification, Clusters, Masses, Spiculate lesions, Asymmetry and Architectural distortions in the mammograms. But the prediction of the cancer levels needs more research light. For the determination of the higher level of accuracy and prediction, the proposed algorithm called Enhanced Gray Scale Adaptive Method (EGAM) which works on the principle of combination of K-GLCM and Extreme Fuzzy Learning Machines (EFLM). The proposed algorithm has achieved 99% accuracy and less computation time in terms of classification, detection and prediction when compared with the existing intelligent algorithms.     

磁共振T1增强扫描对复杂性肛瘘诊断及分型的价值

目的：探讨磁共振T1增强扫描对复杂性肛瘘的诊断价值。方法：选取2015年2月至2018年1月温州医科大学附属第三医院和平阳县人民医院经手术证实的复杂性肛瘘患者43例,全部行T1加权成像（T1WI）、T2加权成像压脂（T2WI FS）和T1加权成像压脂（T1WI FS）增强序列扫描,与手术结果进行对照,比较3种序列对肛瘘内口、主管、支管的显示情况。结果：43例患者中,手术共发现内口67个,主瘘管63个,支管47个。T1WI发现内口10个（占14.93%）、主瘘管34个（占53.97%）,支管10个（占21.28%）。T2WI FS发现内口52个（占77.61%）、主瘘管55个（占87.30%）、支管40个（占85.11%）。T1WI FS增强共发现内口61个（占91.04%）、主瘘管60个（占95.24%）、支管42个（占89.36%）。TIWI FS增强较T2WI FS显示内口的准确性高（P＜0.05）,在显示主瘘管、支管的准确性方面两序列差异无统计学意义（P＞0.05）。T1WI FS增强显示内口、主瘘管、支管的准确性均较T1WI高,差异具有统计学意义（P＜0.05）。T1WI FS增强在Parks分型上与手术结果一致性高（Kappa=0.926）,明显高于T2WI FS与T1WI序列（Kappa=0.786,Kappa=0.467）。结论：T1WI FS增强扫描序列可作为肛瘘MR检查常规T1、T2序列的重要补充,尤其是术前对内口的定位及分型评估中具有重要的指导价值。     

Automated Diagnosis of Diabetes Using Heart Rate Variability Signals

An automated diagnostic system for diabetes mellitus (DM), from heart rate variability (HRV) measures, using feed forward neural network has been developed. Changes in autonomic nervous system activity caused by DM are quantified by means of time domain and frequency domain analysis of HRV. Electrocardiograms of 70 DM patients and 65 healthy volunteers were recorded. Nine time domain measures—standard deviation of all NN intervals, square root of mean of sum of squares of differences between adjacent NN interval (RMSSD), number of adjacent NN intervals differing more than 50 ms. (NN50 count), percentage of NN50 count, R-R triangular index, triangular interpolation of NN intervals (TINN), standard deviation of the mean heart rate, mean R-R interval and mean heart rate—were used as the input features to the neural network. This diagnostic system classifies DM patients and normal volunteers from morphologically identical ECGs. Diagnostic results show that the system is performing well with an accuracy of 93.08%, specificity of 96.92% and sensitivity of 89.23%.     

基于极值随机森林的慢性胃炎中医证候分类

大多数机器学习算法能得到较好的分类效果,但模型却无法解释;而随机森林等模型有良好的可解释性,却无法处理中医数据中兼证的情况。本文利用极值随机森林算法对慢性胃炎中医数据进行证候分类研究,其中决策树的叶节点能输出多个标签,通过加权机制综合分量来处理兼证问题。与已有多标记学习算法和C4.5、CART等基于决策树的算法进行比较,实验结果表明,极值随机森林算法无论在6个证型的分类准确率上,还是在多标记评价指标上都具有更好的效果,而且模型中得到的规则基本符合中医理论。     

Support Vector Machine Based Diagnostic System for Breast Cancer Using Swarm Intelligence

Breast cancer is becoming a leading cause of death among women in the whole world, meanwhile, it is confirmed that the early detection and accurate diagnosis of this disease can ensure a long survival of the patients. In this paper, a swarm intelligence technique based support vector machine classifier (PSO_SVM) is proposed for breast cancer diagnosis. In the proposed PSO-SVM, the issue of model selection and feature selection in SVM is simultaneously solved under particle swarm (PSO optimization) framework. A weighted function is adopted to design the objective function of PSO, which takes into account the average accuracy rates of SVM (ACC), the number of support vectors (SVs) and the selected features simultaneously. Furthermore, time varying acceleration coefficients (TVAC) and inertia weight (TVIW) are employed to efficiently control the local and global search in PSO algorithm. The effectiveness of PSO-SVM has been rigorously evaluated against the Wisconsin Breast Cancer Dataset (WBCD), which is commonly used among researchers who use machine learning methods for breast cancer diagnosis. The proposed system is compared with the grid search method with feature selection by F-score. The experimental results demonstrate that the proposed approach not only obtains much more appropriate model parameters and discriminative feature subset, but also needs smaller set of SVs for training, giving high predictive accuracy. In addition, Compared to the existing methods in previous studies, the proposed system can also be regarded as a promising success with the excellent classification accuracy of 99.3% via 10-fold cross validation (CV) analysis. Moreover, a combination of five informative features is identified, which might provide important insights to the nature of the breast cancer disease and give an important clue for the physicians to take a closer attention. We believe the promising result can ensure that the physicians make very accurate diagnostic decision in clinical breast cancer diagnosis.     

A New Approach to Diagnosing of Importance Degree of Obstructive Sleep Apnea Syndrome: Pairwise AIRS and Fuzzy-AIRS Classifiers

Artificial Immune Recognition System (AIRS) classifier algorithm is robust and effective in medical dataset classification applications such as breast cancer, heart disease, diabetes diagnosis etc. In our previous work, we have proposed a new resource allocation mechanism called fuzzy resource allocation in AIRS algorithm both to improve the classification accuracy and to decrease the computation time in classification process. Here, AIRS and Fuzzy-AIRS classifier algorithms and one against all approach have been combined to increase the classification accuracy of obstructive sleep apnea syndrome (OSAS) that is an important disease that influences both the right and the left cardiac ventricle. The OSAS dataset consists of four classes including of normal (25 subjects), mild OSAS (AHI (Apnea and Hypoapnea Index) =5-15 and 14 subjects), moderate OSAS (AHI < 15-30 and 18 subjects), and serious OSAS (AHI > 30 and 26 subjects). In the extracting of features that is characterized the OSAS disease, the clinical features obtained from Polysomnography used diagnostic tool for obstructive sleep apnea in patients clinically suspected of suffering from this disease have been used. The used clinical features are Arousals Index (ARI), Apnea and Hypoapnea Index (AHI), SaO2 minimum value in stage of REM, and Percent Sleep Time (PST) in stage of SaO2 intervals bigger than 89%. Even though AIRS and Fuzzy-AIRS classifiers have been used in the classifying multi-class problems, theirs classification performances are low in the case of multi-class classification problems. Therefore, we have used two classes in AIRS and Fuzzy-AIRS classifiers by means of one against all approach instead of four classes comprising the healthy subjects, mild OSAS, moderate OSAS, and serious OSAS. We have applied the AIRS, Fuzzy-AIRS, AIRS with one against all approach (Pairwise AIRS), and Fuzzy-AIRS with one against all approach (Pairwise Fuzzy-AIRS) to OSAS dataset. The obtained classification accuracies are 63.41%, 63.41%, 87.19%, and 84.14% using the above methods for 200 resources, respectively. These results show that the best method for diagnosis of OSAS is the combination of AIRS and one against all approach (Pairwise AIRS).     

Breast Cancer Classification Based on Advanced Multi Dimensional Fuzzy Neural Network

Breast cancer is the cause of the most common cancer death in women. Early detection of the breast cancer is an effective method to reduce mortality. Fuzzy Neural Networks (FNN) comprises an integration of the merits of neural and fuzzy approaches, enabling one to build more intelligent decision-making systems. But increasing the number of inputs causes exponential growth in the number of parameters in Fuzzy Neural Networks (FNN) and computational complexity increases accordingly. This phenomenon is named as ??curse of dimensionality??. The Hierarchical Fuzzy Neural Network (HFNN) and the Fuzzy Gaussian Potential Neural Network (FGPNN) are utilized to deal this problem. In this study, the HFNN and FGPNN by using new training algorithm, are applied to the Wisconsin Breast Cancer Database to classify breast cancer into two groups; benign and malignant lesions. The HFNN consists of hierarchically connected low-dimensional fuzzy neural networks. It can use fewer rules and parameters to model nonlinear system. Moreover, the FGPNN consists of Gaussian Potential Function (GPF) used in the antecedent as the membership function. When the number of inputs increases in FGPNN, the number of fuzzy rules does not increase. The performance of HFNN and FGPNN are evaluated and compared with FNN. Simulation results show the effectiveness of these methods even with less rules and parameters in performance result. These methods maintain the accuracy of original fuzzy neural system and have high interpretability by human in diagnosis of breast cancer.     

一种基于噪声分类的语音增强方法

为了提高噪声估计的准确性,改进语音增强方法性能,在改进的最小控制递归平均算法（Improved Minima Controlled Recursive Averaging, IMCRA）的基础上提出了一种基于噪声分类的语音增强方法。该方法首先对含噪语音进行噪声类型的判断,然后根据判定的噪声类型选取相应的最优参数进行噪声估计,最后采用最优修正的对数谱幅度语音估计计算增强后的语音。该方法相对于传统IMCRA算法,在语音信号的还原和背景噪声的抑制两方面都有较好的性能。     

基于核谱回归与随机森林的脑电情感识别

利用DEAP情感数据库研究脑电的情感识别问题。首先,使用聚类算法确定情感状态的目标类别;然后,比较了两种不同的特征提取方法：一种是小波变换,另一种是非线性动力学,并研究了基线特征对情感分类效果的影响;最后,研究了5种特征降维方法对分类性能的影响,同时比较了4种不同分类器的性能,包括K-最近邻（KNN）、朴素贝叶斯（NB）、支持向量机（SVM）和随机森林（RF）。研究结果表明,核谱回归（KSR）降维方法和随机森林分类器的组合对情感状态的分类效果最好。通过对脑区与情感关系的研究发现,只使用部分脑区的少量电极也可以达到90%的分类准确度,这些电极主要分布在额叶皮层。     

A Mixture of Experts Network Structure for Breast Cancer Diagnosis

Mixture of experts (ME) is a modular neural network architecture for supervised learning. This paper illustrates the use of ME network structure to guide diagnosing of breast cancer. Expectation-maximization (EM) algorithm was used for training the ME so that the learning process is decoupled in a manner that fits well with the modular structure. Diagnosis tasks are among the most interesting activities in which to implement intelligent systems. Specifically, diagnosis is an attempt to accurately forecast the outcome of a specific situation, using as input information obtained from a concrete set of variables that potentially describe the situation. The ME network structure was implemented for breast cancer diagnosis using the attributes of each record in the Wisconsin breast cancer database. To improve diagnostic accuracy, the outputs of expert networks were combined by a gating network simultaneously trained in order to stochastically select the expert that is performing the best at solving the problem. For the Wisconsin breast cancer diagnosis problem, the obtained total classification accuracy by the ME network structure was 98.85%. The ME network structure achieved accuracy rates which were higher than that of the stand-alone neural network models.     

Automatic classification of mammography reports by BI-RADS breast tissue composition class

Because breast tissue composition partially predicts breast cancer risk, classification of mammography reports by breast tissue composition is important from both a scientific and clinical perspective. A method is presented for using the unstructured text of mammography reports to classify them into BI-RADS breast tissue composition categories. An algorithm that uses regular expressions to automatically determine BI-RADS breast tissue composition classes for unstructured mammography reports was developed. The algorithm assigns each report to a single BI-RADS composition class: 'fatty', 'fibroglandular', 'heterogeneously dense', 'dense', or 'unspecified'. We evaluated its performance on mammography reports from two different institutions. The method achieves >99% classification accuracy on a test set of reports from the Marshfield Clinic (Wisconsin) and Stanford University. Since large-scale studies of breast cancer rely heavily on breast tissue composition information, this method could facilitate this research by helping mine large datasets to correlate breast composition with other covariates.     

Active learning for clinical text classification: is it better than random sampling?

Objective

This study explores active learning algorithms as a way to reduce the requirements for large training sets in medical text classification tasks.

Design

Three existing active learning algorithms (distance-based (DIST), diversity-based (DIV), and a combination of both (CMB)) were used to classify text from five datasets. The performance of these algorithms was compared to that of passive learning on the five datasets. We then conducted a novel investigation of the interaction between dataset characteristics and the performance results.

Measurements

Classification accuracy and area under receiver operating characteristics (ROC) curves for each algorithm at different sample sizes were generated. The performance of active learning algorithms was compared with that of passive learning using a weighted mean of paired differences. To determine why the performance varies on different datasets, we measured the diversity and uncertainty of each dataset using relative entropy and correlated the results with the performance differences.

Results

The DIST and CMB algorithms performed better than passive learning. With a statistical significance level set at 0.05, DIST outperformed passive learning in all five datasets, while CMB was found to be better than passive learning in four datasets. We found strong correlations between the dataset diversity and the DIV performance, as well as the dataset uncertainty and the performance of the DIST algorithm.

Conclusion

For medical text classification, appropriate active learning algorithms can yield performance comparable to that of passive learning with considerably smaller training sets. In particular, our results suggest that DIV performs better on data with higher diversity and DIST on data with lower uncertainty.     

Modified Mixture of Experts for Diabetes Diagnosis

Diagnosis tasks are among the most interesting activities in which to implement intelligent systems. The major objective of the paper is to be a guide for the readers, who want to develop an automated decision support system for detection of diabetics and subjects having risk factors of diabetes. The purpose was to determine an optimum classification scheme with high diagnostic accuracy for this problem. Several different classification algorithms were tested and their performances in detection of diabetics were compared. The performance of the classification algorithms was illustrated on the Pima Indians diabetes data set. The present research demonstrated that the modified mixture of experts (MME) achieved diagnostic accuracies which were higher than that of the other automated diagnostic systems.     

A New Approach: Role of Data Mining in Prediction of Survival of Burn Patients

The prediction of burn patient survivability is a difficult problem to investigate till present times. In present study a prediction Model for patients with burns was built, and its capability to accurately predict the survivability was assessed. We have compared different data mining techniques to asses the performance of various algorithms based on the different measures used in the analysis of information pertaining to medical domain. Obtained results were evaluated for correctness with the help of registered medical practitioners. The dataset was collected from SRT (Swami Ramanand Tirth) Hospital in India, which is one of the Asia’s largest rural hospitals. Dataset contains records of 180 patients mainly suffering from burn injuries collected during period from the year 2002 to 2006. Features contain patients’ age, sex and percentage of burn received for eight different parts of the body. Prediction models have been developed through rigorous comparative study of important and relevant data mining classification techniques namely, navie bayes, decision tree, support vector machine and back propagation. Performance comparison was also carried out for measuring unbiased estimate of the prediction models using 10-fold cross-validation method. Using the analysis of obtained results, we show that Navie bayes is the best predictor with an accuracy of 97.78% on the holdout samples, further, both the decision tree and support vector machine (SVM) techniques demonstrated an accuracy of 96.12%, and back propagation technique resulted in achieving accuracy of 95%.     

Statistical Analysis of Textural Features for Improved Classification of Oral Histopathological Images

The objective of this paper is to provide an improved technique, which can assist oncopathologists in correct screening of oral precancerous conditions specially oral submucous fibrosis (OSF) with significant accuracy on the basis of collagen fibres in the sub-epithelial connective tissue. The proposed scheme is composed of collagen fibres segmentation, its textural feature extraction and selection, screening perfomance enhancement under Gaussian transformation and finally classification. In this study, collagen fibres are segmented on R,G,B color channels using back-probagation neural network from 60 normal and 59 OSF histological images followed by histogram specification for reducing the stain intensity variation. Henceforth, textural features of collgen area are extracted using fractal approaches viz., differential box counting and brownian motion curve . Feature selection is done using Kullback–Leibler (KL) divergence criterion and the screening performance is evaluated based on various statistical tests to conform Gaussian nature. Here, the screening performance is enhanced under Gaussian transformation of the non-Gaussian features using hybrid distribution. Moreover, the routine screening is designed based on two statistical classifiers viz., Bayesian classification and support vector machines (SVM) to classify normal and OSF. It is observed that SVM with linear kernel function provides better classification accuracy (91.64%) as compared to Bayesian classifier. The addition of fractal features of collagen under Gaussian transformation improves Bayesian classifier’s performance from 80.69% to 90.75%. Results are here studied and discussed.     

术前左室射血分数及短缩分数对冠状动脉旁路术后恶性心律失常的预测价值

为了评价冠状动脉旁路术 (CABG)术前左室射血分数 (LVEF)及左室短缩分数 (LVFS)对术后室性心律失常 (VA)预测的准确性 ,采用术前及术后 2周心脏彩超EF、FS值 (面积长轴法 )、心室晚电位 (VLP)、心肌酶、持续心电监测的方法 ,对我院 1 5 0例行CABG术的患者进行分析。结果 :1 )术前心肌梗死 (MI)、室壁瘤、VA及VLP阳性患者EF、FS值明显减低 ;2 )术前左心功能不全 (LVD)患者术后EF、FS值明显改善 ;3 )术前LVD、VA、VLP阳性及室壁瘤患者术后VA发生率明显高于其他患者。提示 :1 )面积长轴法EF、FS值是反映左心功能的敏感指标 ;2 )术前LVD患者术后短期左心功能明显好转 ,获益最大 ;3 )非LVD患者术后因心肌顿抑导致近期心功能暂时下降 ;4 )EF≤ 4 0 %和(或 )FS≤ 2 4 %是预测术后VA的独立指标 ,FS较EF更能准确地反映心脏收缩功能 ;5 )LVD、VLP、室壁瘤等综合指标分析有助于提高对术后VA预期的敏感性、特异性和准确性