An improved algorithm for the automatic detection and characterization of slow eye movements

Slow eye movements (SEMs) are typical of drowsy wakefulness and light sleep. SEMs still lack of systematic physical characterization. We present a new algorithm, which substantially improves our previous one, for the automatic detection of SEMs from the electro-oculogram (EOG) and extraction of SEMs physical parameters. The algorithm utilizes discrete wavelet decomposition of the EOG to implement a Bayes classifier that identifies intervals of slow ocular activity; each slow activity interval is segmented into single SEMs via a template matching method. Parameters of amplitude, duration, velocity are automatically extracted from each detected SEM. The algorithm was trained and validated on sleep onsets and offsets of 20 EOG recordings visually inspected by an expert. Performances were assessed in terms of correctly identified slow activity epochs (sensitivity: 85.12%; specificity: 82.81%), correctly segmented single SEMs (89.08%), and time misalignment (0.49 s) between the automatically and visually identified SEMs. The algorithm proved reliable even in whole sleep (sensitivity: 83.40%; specificity: 72.08% in identifying slow activity epochs; correctly segmented SEMs: 93.24%; time misalignment: 0.49 s). The algorithm, being able to objectively characterize single SEMs, may be a valuable tool to improve knowledge of normal and pathological sleep.     

Ranking and combining multiple predictors without labeled data

In a broad range of classification and decision-making problems, one is given the advice or predictions of several classifiers, of unknown reliability, over multiple questions or queries. This scenario is different from the standard supervised setting, where each classifier’s accuracy can be assessed using available labeled data, and raises two questions: Given only the predictions of several classifiers over a large set of unlabeled test data, is it possible to (i) reliably rank them and (ii) construct a metaclassifier more accurate than most classifiers in the ensemble? Here we present a spectral approach to address these questions. First, assuming conditional independence between classifiers, we show that the off-diagonal entries of their covariance matrix correspond to a rank-one matrix. Moreover, the classifiers can be ranked using the leading eigenvector of this covariance matrix, because its entries are proportional to their balanced accuracies. Second, via a linear approximation to the maximum likelihood estimator, we derive the Spectral Meta-Learner (SML), an unsupervised ensemble classifier whose weights are equal to these eigenvector entries. On both simulated and real data, SML typically achieves a higher accuracy than most classifiers in the ensemble and can provide a better starting point than majority voting for estimating the maximum likelihood solution. Furthermore, SML is robust to the presence of small malicious groups of classifiers designed to veer the ensemble prediction away from the (unknown) ground truth.Every day, multiple decisions are made based on input and suggestions from several sources, either algorithms or advisers, of unknown reliability. Investment companies handle their portfolios by combining reports from several analysts, each providing recommendations on buying, selling, or holding multiple stocks (1, 2). Central banks combine surveys of several professional forecasters to monitor rates of inflation, real gross domestic product growth, and unemployment (3–6). Biologists study the genomic binding locations of proteins by combining or ranking the predictions of several peak detection algorithms applied to large-scale genomics data (7). Physician tumor boards convene a number of experts from different disciplines to discuss patients whose diseases pose diagnostic and therapeutic challenges (8). Peer-review panels discuss multiple grant applications and make recommendations to fund or reject them (9). The examples above describe scenarios in which several human advisers or algorithms provide their predictions or answers to a list of queries or questions. A key challenge is to improve decision making by combining these multiple predictions of unknown reliability. Automating this process of combining multiple predictors is an active field of research in decision science (cci.mit.edu/research), medicine (10), business (refs. 11 and 12 and www.kaggle.com/competitions), and government (www.iarpa.gov/Programs/ia/ACE/ace.html and www.goodjudgmentproject.com), as well as in statistics and machine learning.Such scenarios, whereby advisers of unknown reliability provide potentially conflicting opinions, or propose to take opposite actions, raise several interesting questions. How should the decision maker proceed to identify who, among the advisers, is the most reliable? Moreover, is it possible for the decision maker to cleverly combine the collection of answers from all of the advisers and provide even more accurate answers?In statistical terms, the first question corresponds to the problem of estimating prediction performances of preconstructed classifiers (e.g., the advisers) in the absence of class labels. Namely, each classifier was constructed independently on a potentially different training dataset (e.g., each adviser trained on his/her own using possibly different sources of information), yet they are all being applied to the same new test data (e.g., list of queries) for which labels are not available, either because they are expensive to obtain or because they will only be available in the future, after the decision has been made. In addition, the accuracy of each classifier on its own training data is unknown. This scenario is markedly different from the standard supervised setting in machine learning and statistics. There, classifiers are typically trained on the same labeled data and can be ranked, for example, by comparing their empirical accuracy on a common labeled validation set. In this paper we show that under standard assumptions of independence between classifier errors their unknown performances can still be ranked even in the absence of labeled data.The second question raised above corresponds to the problem of combining predictions of preconstructed classifiers to form a metaclassifier with improved prediction performance. This problem arises in many fields, including combination of forecasts in decision science and crowdsourcing in machine learning, which have each derived different approaches to address it. If we had external knowledge or historical data to assess the reliability of the available classifiers we could use well-established solutions relying on panels of experts or forecast combinations (11–14). In our problem such knowledge is not always available and thus these solutions are in general not applicable. The oldest solution that does not require additional information is majority voting, whereby the predicted class label is determined by a rule of majority, with all advisers assigned the same weight. More recently, iterative likelihood maximization procedures, pioneered by Dawid and Skene (15), have been proposed, in particular in crowdsourcing applications (16–23). Owing to the nonconvexity of the likelihood function, these techniques often converge only to a local, rather than global, maximum and require careful initialization. Furthermore, there are typically no guarantees on the quality of the resulting solution.In this paper we address these questions via a spectral analysis that yields four major insights:

1. Under standard assumptions of independence between classifier errors, in the limit of an infinite test set, the off-diagonal entries of the population covariance matrix of the classifiers correspond to a rank-one matrix.
2. The entries of the leading eigenvector of this rank-one matrix are proportional to the balanced accuracies of the classifiers. Thus, a spectral decomposition of this rank-one matrix provides a computationally efficient approach to rank the performances of an ensemble of classifiers.
3. A linear approximation of the maximum likelihood estimator yields an ensemble learner whose weights are proportional to the entries of this eigenvector. This represents an efficient, easily constructed, unsupervised ensemble learner, which we term Spectral Meta-Learner (SML).
4. An interest group of conspiring classifiers (a cartel) that maliciously attempts to veer the overall ensemble solution away from the (unknown) ground truth leads to a rank-two covariance matrix. Furthermore, in contrast to majority voting, SML is robust to the presence of a small-enough cartel whose members are unknown.

In addition, we demonstrate the advantages of spectral approaches based on these insights, using both simulated and real-world datasets. When the independence assumptions hold approximately, SML is typically better than most classifiers in the ensemble and their majority vote, achieving results comparable to the maximum likelihood estimator (MLE). Empirically, we find SML to be a better starting point for computing the MLE that consistently leads to improved performance. Finally, spectral approaches are also robust to cartels and therefore helpful in analyzing surveys where a biased subgroup of advisers (a cartel) may have corrupted the data.     

A weighted rule based method for predicting malignancy of pulmonary nodules by nodule characteristics

Predicting malignancy of solitary pulmonary nodules from computer tomography scans is a difficult and important problem in the diagnosis of lung cancer. This paper investigates the contribution of nodule characteristics in the prediction of malignancy. Using data from Lung Image Database Consortium (LIDC) database, we propose a weighted rule based classification approach for predicting malignancy of pulmonary nodules. LIDC database contains CT scans of nodules and information about nodule characteristics evaluated by multiple annotators. In the first step of our method, votes for nodule characteristics are obtained from ensemble classifiers by using image features. In the second step, votes and rules obtained from radiologist evaluations are used by a weighted rule based method to predict malignancy. The rule based method is constructed by using radiologist evaluations on previous cases. Correlations between malignancy and other nodule characteristics and agreement ratio of radiologists are considered in rule evaluation. To handle the unbalanced nature of LIDC, ensemble classifiers and data balancing methods are used. The proposed approach is compared with the classification methods trained on image features. Classification accuracy, specificity and sensitivity of classifiers are measured. The experimental results show that using nodule characteristics for malignancy prediction can improve classification results.     

基于近红外光谱技术天麻的产地区分

天麻在《神农本草经》中列为上品,其主要分布在我国西南地区,不同产地质量差别较大,野生与栽培品价格差别尤为悬殊。用传统性状鉴别方法难以客观准确区分天麻的质量,故本文采用近红外光谱技术结合模式识别方法,建立能够快速准确区分不同产地和不同生长、栽培方式的天麻药材的方法。结果表明,利用近红外光谱技术结合多类分类算法,可以实现天麻产地和栽培方式的快速判别,其交叉验证判别准确率分别达到94.3%和96.4%,且所建多类分类模型中光谱数据未经预处理。因此,本研究所建立的天麻产地和质量的快速判定方法可以推广为现场应用。     

Emotion recognition from physiological signals

Emotion recognition is one of the great challenges in human–human and human–computer interaction. Accurate emotion recognition would allow computers to recognize human emotions and therefore react accordingly. In this paper, an approach for emotion recognition based on physiological signals is proposed. Six basic emotions: joy, sadness, fear, disgust, neutrality and amusement are analysed using physiological signals. These emotions are induced through the presentation of International Affecting Picture System (IAPS) pictures to the subjects. The physiological signals of interest in this analysis are: electromyogram signal (EMG), respiratory volume (RV), skin temperature (SKT), skin conductance (SKC), blood volume pulse (BVP) and heart rate (HR). These are selected to extract characteristic parameters, which will be used for classifying the emotions. The SVM (support vector machine) technique is used for classifying these parameters. The experimental results show that the proposed methodology provides in general a recognition rate of 85% for different emotional states.     

Abnormal loading and functional deficits are present in both limbs before and after unilateral knee arthroplasty

Unilateral knee replacement is often followed by a contralateral replacement in time and the biomechanics of the other knee before and after knee replacement remains poorly understood. The aim of this paper is to distinguish the features of arthritic gait in the affected and unaffected legs relative to a normal population and to assess the objective recovery of gait function post-operatively, with the aim of defining patients at risk of poor post-operative function. Twenty patients with severe knee OA but no pain or deformity in any other lower limb joint were compared to twenty healthy subjects of the same age. Gait analysis was performed and quadriceps and hamstrings co-contraction was measured. Fifteen subjects returned 1 year following knee arthroplasty. Moments and impulses were calculated, principal component analysis was used to analyse the waveforms and a classification technique (the Cardiff Classifier) was used to select the most discriminant data and define functional performance. Comparing pre-operative function to healthy function, classification accuracies for the affected and unaffected knees were 95% and 92.5% respectively. Post-operatively, the affected limb returned to the normal half of the classifier in 8 patients, and 7 of those patients returned to normal function in the unaffected limb. Recovery of normal gait could be correctly predicted 13 out of 15 times at the affected knee, and 12 out of 15 times at the unaffected knee based on pre-operative gait function. Focused rehabilitation prior to surgery may be beneficial to optimise outcomes and protect the other joints following knee arthroplasty.     

计算机辅助系统在诊断乳腺良恶性肿瘤中的应用

目的:探讨计算机辅助诊断系统在良恶性肿瘤检测与特征提取基础上的分类对于乳腺肿瘤的诊断价值。方法:回顾性分析乳腺超声检查发现肿瘤且经过病理学证实的617例患者影像资料,采用手工提取的方式得到乳腺超声图像的感兴趣区域及病灶轮廓,再利用方向梯度直方图（HOG）、局部二值模式（LBP）和灰度共生矩阵（GLCM）3个特征进行乳腺肿瘤的良恶性病变真假阳性检测;最后用受试者操作特征曲线（ROC）分别分析每个特征对于两类病变判别的诊断性能和应用所有特征集合的分类诊断性能。结果:多特征融合方法的各项诊断效能及ROC曲线下面积（AUC）值均优于单特征LBP、HOG、GLCM（P值均<0.05）。与人工诊断相比,多特征融合的敏感性无显著差异,但特异度显著升高达98.57%（Z值=2.25, P<0.05）,同时AUC值为0.985,显著优于人工诊断的0.910（Z值=1.99, P<0.05）。结论:计算机辅助系统乳腺超声肿瘤良恶性检测的算法是有效的,能够对乳腺癌鉴别诊断提供有益的参考。