Search

This article presents a~study of the approaches in the state-of-the-art in the field of pathological speech signal analysis with a~special focus on parametrization techniques. It provides a~description of 92 speech features where some of them are already widely used in this field of science and some of them have not been tried yet (they come from different areas of speech signal processing like speech recognition or coding). As an original contribution, this work introduces 36 completely new pathological voice measures based on modulation spectra, inferior colliculus coefficients, bicepstrum, sample and approximate entropy and empirical mode decomposition. The significance of these features was tested on 3 (English, Spanish and Czech) pathological voice databases with respect to classification accuracy, sensitivity and specificity. To our best knowledge the introduced approach based on complex feature extraction and robust testing outperformed all works that have been published already in this field. The results (accuracy, sensitivity and specificity equal to $100.0\pm0.0\,\%$) are discussable in the case of Massachusetts Eye and Ear Infirmary (MEEI) database because of its limitation related to a~length of sustained vowels, however in the case of Pr{\'i}ncipe de Asturias (PdA) Hospital in Alcal{\'a} de Henares of Madrid database we made improvements in classification accuracy ($82.1\pm3.3\,\%$) and specificity ($83.8\pm5.1\,\%$) when considering a~single-classifier approach. Hopefully, large improvements may be achieved in the case of Czech Parkinsonian Speech Database (PARCZ), which are discussed in this work as well. All the features introduced in this work were identified by Mann-Whitney~U test as significant ($p < 0.05$) when processing at least one of the mentioned databases. The largest discriminative power from these proposed features has a~cepstral peak prominence extracted from the first intrinsic mode function ($p = 6.9443\cdot10^{-32}$) which means, that among all newly designed features those that quantify especially hoarseness or breathiness are good candidates for pathological speech identification. The article also mentions some ideas for the future work in the field of pathological speech signal analysis that can be valuable especially under the clinical point of view.

This paper presents a study of the approaches in the state-of-the-art in the field of pathological speech signal analysis with a special focus on parametrization techniques. It provides a description of 92 speech features where some of them are already widely used in this field of science and some of them have not been tried yet (they come from different areas of speech signal processing like speech recognition or coding). As an original contribution, this work introduces 36 completely new pathological voice measures based on modulation spectra, inferior colliculus coefficients, bicepstrum, sample and approximate entropy and empirical mode decomposition. The significance of these features was tested on 3 (English, Spanish and Czech) pathological voice databases with respect to classification accuracy, sensitivity and specificity., Comment: 41 pages, published in Neurocomputing, Volume 167, 2015, Pages 94-111, ISSN 0925-2312

In this paper, we evaluate the contribution of different handwriting modalities to the diagnosis of Parkinson's disease. We analyse on-surface movement, in-air movement and pressure exerted on the tablet surface. Especially in-air movement and pressure-based features have been rarely taken into account in previous studies. We show that pressure and in-air movement also possess information that is relevant for the diagnosis of Parkinson's Disease (PD) from handwriting. In addition to the conventional kinematic and spatio-temporal features, we present a group of the novel features based on entropy and empirical mode decomposition of the handwriting signal. The presented results indicate that handwriting can be used as biomarker for PD providing classification performance around 89% area under the ROC curve (AUC) for PD classification., Comment: The work was published by IEEE

Hypokinetic dysarthria (HD) and freezing of gait (FOG) are both axial symptoms that occur in patients with Parkinson’s disease (PD). It is assumed they have some common pathophysiological mechanisms and therefore that speech disorders in PD can predict FOG deficits within the horizon of some years. The aim of this study is to employ a complex quantitative analysis of the phonation, articulation and prosody in PD patients in order to identify the relationship between HD and FOG, and establish a mathematical model that would predict FOG deficits using acoustic analysis at baseline. We enrolled 75 PD patients who were assessed by 6 clinical scales including the Freezing of Gait Questionnaire (FOG–Q). We subsequently extracted 19 acoustic measures quantifying speech disorders in the fields of phonation, articulation and prosody. To identify the relationship between HD and FOG, we performed a partial correlation analysis. Finally, based on the selected acoustic measures, we trained regression models to predict the change in FOG during a 2-year follow-up. We identified significant correlations between FOG–Q scores and the acoustic measures based on formant frequencies (quantifying the movement of the tongue and jaw) and speech rate. Using the regression models, we were able to predict a change in particular FOG–Q scores with an error of between 7.4 and 17.0 %. This study is suggesting that FOG in patients with PD is mainly linked to improper articulation, a disturbed speech rate and to intelligibility. We have also proved that the acoustic analysis of HD at the baseline can be used as a predictor of the FOG deficit during 2 years of follow-up. This knowledge enables researchers to introduce new cognitive systems that predict gait difficulties in PD patients.

Automatic objective non-invasive detection of pathological voice based on computerized analysis of acoustic signals can play an important role in early diagnosis, progression tracking, and even effective treatment of pathological voices. In search towards such a robust voice pathology detection system, we investigated three distinct classifiers within supervised learning and anomaly detection paradigms. We conducted a set of experiments using a variety of input data such as raw waveforms, spectrograms, mel-frequency cepstral coefficients (MFCC), and conventional acoustic (dysphonic) features (AF). In comparison with previously published works, this article is the first to utilize combination of four different databases comprising normophonic and pathological recordings of sustained phonation of the vowel /a/ unrestricted to a subset of vocal pathologies. Furthermore, to our best knowledge, this article is the first to explore gradient-boosted trees and deep learning for this application. The following best classification performances measured by F1 score on dedicated test set were achieved: XGBoost (0.733) using AF and MFCC, DenseNet (0.621) using MFCC, and Isolation Forest (0.610) using AF. Even though these results are of exploratory character, conducted experiments do show promising potential of gradient boosting and deep learning methods to robustly detect voice pathologies.

Developmental dysgraphia, being observed among 10–30% of school-aged children, is a disturbance or difficulty in the production of written language that has to do with the mechanics of writing. The objective of this study is to propose a method that can be used for automated diagnosis of this disorder, as well as for estimation of difficulty level as determined by the handwriting proficiency screening questionnaire. We used a digitizing tablet to acquire handwriting and consequently employed a complex parameterization in order to quantify its kinematic aspects and hidden complexities. We also introduced a simple intrawriter normalization that increased dysgraphia discrimination and HPSQ estimation accuracies. Using a random forest classifier, we reached 96% sensitivity and specificity, while in the case of automated rating by the HPSQ total score, we reached 10% estimation error. This study proves that digital parameterization of pressure and altitude/tilt patterns in children with dysgraphia can be used for preliminary diagnosis of this writing disorder.

Convolutional neural network (CNN) architectures have originated and revolutionized machine learning for images. In order to take advantage of CNNs in predictive modeling with audio data, standard FFT-based signal processing methods are often applied to convert the raw audio waveforms into an image-like representations (e.g. spectrograms). Even though conventional images and spectrograms differ in their feature properties, this kind of pre-processing reduces the amount of training data necessary for successful training. In this contribution we investigate how input and target representations interplay with the amount of available training data in a music information retrieval setting. We compare the standard mel-spectrogram inputs with a newly proposed representation, called Mel scattering. Furthermore, we investigate the impact of additional target data representations by using an augmented target loss function which incorporates unused available information. We observe that all proposed methods outperform the standard mel-transform representation when using a limited data set and discuss their strengths and limitations. The source code for reproducibility of our experiments as well as intermediate results and model checkpoints are available in an online repository., Comment: 13 pages, 3 figures, 2 tables. Repository for reproducibility: https://gitlab.com/hararticles/gs-ms-mt/. Keywords: audio, CNN, limited data, Mel scattering, mel-spectrogram, augmented target loss function. Rewritten and restructured after peer revision. Recomputed and added new experiments and visualizations. Changed the presentation of the results

Handwriting difficulties (HD) affects some of the school-aged children and its current prevalence rate is between 5–34%. Children at primary schools have to face rising cognitive demands that the handwriting represents, and some of them are not able to do so. As a result, they tend to make mistakes and their written product is dysfluent and has poor legibility. HD can also lead them to lower self-esteem, learning difficulties and ultimately to less academic achievements. For this reason occupational therapists are trying to identify HD through examination as early as possible. We extracted online handwriting signals of children using digitizing tablets. Handwriting Proficiency Screening Questionnaire for Children (HPSQ–C) was used to score severity of HD in children’s written product. To advance current computerized analysis of online handwriting, we employed fractional order derivative features (FD) together with conventional measures. We selected significant features for HD identification and utilized correlation analysis together with Mann-Whitney U-test to evaluate their discriminative power. We can conclude that FD-based features bring benefits of more robust quantification of in-air movements as opposed to the conventionally used ones. Finally, we have shown that utilization of FD can be beneficial for computerized assessment of HD but should be further optimized and evaluated with advanced statistical or machine learning methods.

Although graphomotor difficulties (GD) are present in up to 30% of school-aged children, the field of GD diagnosis and assessment is not fully explored and several research gaps can be identified. This study aims to explore the impact of specific elementary graphomotor tasks analysis on the accuracy of computerised diagnosis and assessment of GD. We analysed seven basic graphomotor tasks from 76 children (assessed by special educational counsellors and using the handwriting proficiency screening questionnaire for children HPSQ–C). Employing a differential analysis, we observed that the most discriminative tasks are based on combined loops, sawtooth and small Archimedean spiral drawings. Features with the highest discrimination power quantify kinematics, especially in the vertical projection. Using a multivariate mathematical model, we were able to identify GD with 50% sensitivity and 90% specificity, and to estimate the total score of HPSQ–C with 31% error.

Parkinson’s disease (PD) is a common neurodegenerative disorder with prevalence rate estimated to 1.5 % for people age over 65 years. The majority of PD patients is associated with handwriting abnormalities called PD dysgraphia, which is linked with rigidity and bradykinesia of muscles involved in the handwriting process. One of the effective approaches of quantitative PD dysgraphia analysis is based on online handwriting processing. In the frame of this study we aim to deeply evaluate and optimize advanced PD handwriting quantification based on fractional order derivatives (FD). For this purpose, we used 37 PD patients and 38 healthy controls from the PaHaW (PD handwriting database). The FD based features were employed in classification and regression analysis (using gradient boosted trees), and evaluated in terms of their discrimination power and abilities to assess severity of PD. The results suggest that the most discriminative and descriptive information provide FD based features extracted from a repetitive loop task or a sentence copy task (maximum sensitivity/specificity = 76 %, error in severity assessment = 14 %, error in PD duration estimation = 22 %). Next, we identified two optimal ranges for the order of fractional derivative, α = 0.05–0.45 and α = 0.65–0.80. Finally, we observed that inclusion of pressure, azimuth, and tilt together with kinematic features into mathematical modeling has no influence (positive or negative) on classification performance, however, there was a notable improvement in the estimation of PD duration.

Beat detection systems are widely used in music information retrieval (MIR) research field for the computation of tempo and beat positions in audio signals. One of the most important parts of these systems is the onset detection function. The aim of this study is to introduce an enhancement of a conventional onset detector and employ it in a beat tracking system that could be utilized for an analysis of interpretation and performance changes in string quartets. The enhancement is based on the Teager-Kaiser energy operator (TKEO), which pre-processes input audio signal before spectral flux calculation. The proposed approach is firstly evaluated in terms of ability to estimate global tempo (GT) of a given audio track. Next, the accuracy of the GT estimation is compared with a manually-labelled reference dataset. Then, this system was tested on the first motif of the string quartet database. Results suggest that the TKEO could improve accuracy of the GT estimation. Average deviation from the reference tempo in the string quartet is 8.29%, which slightly improves the conventional methodology, where the deviance is 8.96%. This study has a pilot character and provides some suggestions of the beat tracking system enhancement.

Developmental dysgraphia is a neurodevelopmental disorder present in up to 30% of elementary school pupils. Since it is associated with handwriting difficulties (HD), it has detrimental impact on children’s academic progress, emotional well-being, attitude and behaviour. Nowadays, researchers proposed a new approach of HD assessment utilizing digitizing tablets. I.e. that handwriting of children is quantified by a set of conventional parameters, such as velocity, duration of handwriting, tilt, etc. The aim of this study is to explore a potential of newly designed online handwriting features based on the tunable Q-factor wavelet transform (TQWT) in terms of computerized HD identification. Using a digitizing tablet, we recorded a written paragraph of 97 children who were also assessed by the Handwriting Proficiency Screening Questionnaire for Children (HPSQ–C). We evaluated discrimination power (binary classification) of all parameters using random forest and support vector machine classifiers in combination with sequential floating forward feature selection. Based on the experimental results we observed that the newly designed features outperformed the conventional ones (accuracy = 79.16%, sensitivity = 86.22%, specificity = 73.32%). When considering the combination of all parameters (including the conventional ones) we reached 84.66% classification accuracy (sensitivity = 88.70%, specificity = 82.53%). The most discriminative parameters were based on vertical movement and pressure, which suggests that children with HD were not able to maintain stable force on pen tip and that their vertical movement is less fluent. The new features we introduced go beyond the state-of-the-art and improve discrimination power of the conventional parameters by approximately 20.0%.

Developmental dysgraphia (DD) can negatively influence writing speed, legibility of written product, and even planning and content generation. This could have a detrimental impact on self-expression, and communication in childhood. The goal of this study is to investigate new intra-writer and self-esteem normalization methods (IWN) in direction of improving computerized DD assessment based on the quantitative analysis of online handwriting. For this purpose we enrolled 97 children who wrote a paragraph using a digitizing tablet. Their handwriting proficiency was rated by the Handwriting Proficiency Screening Questionnaire (HPSQ). The handwriting was parametrized using a conventional set of features that were consequently normalized by four newly designed IWN. Based on the results we observed that a stroke-level £2 norm normalization decreased the computerized DD assessment error from 23 % to 18 %. This study proves that the stroke-level IWN has the significant potential in the field of computerized DD diagnosis and assessment.

Determining the place of origin of the musical compositions is a modern area of research in the field of music information retrieval (MIR). The musical interpretation of one piece carries a variety of author's intentions that influence the musical character of the resulting composition. These aspects may include rhythm, dynamics, timbre, or tonality. This paper introduces a novel methodology for determining the place of origin of a music interpretation based on advanced signal processing and machine learning techniques. For this purpose, we collected a database of 35 different interpretations of Leos Janacek's String Quartet No.1, “Kreutzer Sonat”: IV. Con Moto-Adagio. Employing random forests classifier, we achieved classification accuracy over 97 % using features derived from Mel-frequency cepstral coefficients. This paper proves it is possible to use MRI for determining the origin of a music interpretation with very high accuracy.

Hypokinetic dysarthria, which is associated with Parkinson&rsquo, s disease (PD), affects several speech dimensions, including phonation. Although the scientific community has dealt with a quantitative analysis of phonation in PD patients, a complex research revealing probable relations between phonatory features and progress of PD is missing. Therefore, the aim of this study is to explore these relations and model them mathematically to be able to estimate progress of PD during a two-year follow-up. We enrolled 51 PD patients who were assessed by three commonly used clinical scales. In addition, we quantified eight possible phonatory disorders in five vowels. To identify the relationship between baseline phonatory features and changes in clinical scores, we performed a partial correlation analysis. Finally, we trained XGBoost models to predict the changes in clinical scores during a two-year follow-up. For two years, the patients&rsquo, voices became more aperiodic with increased microperturbations of frequency and amplitude. Next, the XGBoost models were able to predict changes in clinical scores with an error in range 11&ndash, 26%. Although we identified some significant correlations between changes in phonatory features and clinical scores, they are less interpretable. This study suggests that it is possible to predict the progress of PD based on the acoustic analysis of phonation. Moreover, it recommends utilizing the sustained vowel /i/ instead of /a/.

Hypokinetic dysarthria (HD) is a frequent symptom of idiopathic Parkinson's disease (PD). Although it is hypothesized its progress is tightly linked with changes in other motor/non-motor features of PD, it has not been proved yet. The aim of this work is to employ acoustic analysis of sustained phonation in order to identify significant correlates between phonatory measures and motor/non-motor deficits in a two-year follow-up study. For this purpose, we repeatedly quantified a sustained vowel/a/in 51 PD patients who were also assessed by 5 common clinical scales. In addition, a multivariate regression model was trained to predict the motor/non-motor deficits in the horizon of two years. Results suggest that mainly instability in vocal folds oscillation increases with the progress of PD and with overall cognitive decline. Based on the acoustic analysis, the change in clinical scores could be predicted with the error in the range of 11.83-19.60 %.

Feature selection is a significant part of many machine learning applications dealing with small-sample and high-dimensional data. Choosing the most important features is an essential step for knowledge discovery in many areas of biomedical informatics. The increased popularity of feature selection methods and their frequent utilisation raise challenging new questions about the interpretability and stability of feature selection techniques. In this study, we compared the behaviour of ten state-of-the-art filter methods for feature selection in terms of their stability, similarity, and influence on prediction performance. All of the experiments were conducted on eight two-class datasets from biomedical areas. While entropy-based feature selection appears to be the most stable, the feature selection techniques yielding the highest prediction performance are minimum redundance maximum relevance method and feature selection based on Bhattacharyya distance. In general, univariate feature selection techniques perform similarly to or even better than more complex multivariate feature selection techniques with high-dimensional datasets. However, with more complex and smaller datasets multivariate methods slightly outperform univariate techniques. HighlightsTen feature selection methods are compared using stability and similarity measures.Univariate FS perform better than multivariate FS for high dimensional datasets.Multivariate FS slightly outperform univariate FS for complex and smaller datasets.Most stable appears to be entropy based FS.FS yielding the highest prediction performance are MRMR and Bhattacharyya distance.

The aim of this work is to propose the fully automated pathological area extraction from multi-parametric 2D MR images of brain. The proposed method is based on multi-resolution symmetry analysis and automatic thresholding. The proposed algorithm first detects the presence of pathology and then starts its extraction. T2 images are used for the presence detection and the multi-contrast MRI is used for the extraction, concretely T2 and FLAIR images. The extraction is based on thresholding, where Otsu's algorithm is used for the automatic determination of the threshold. Since the method is based on symmetry, it works for both axial and coronal planes. In both these planes of healthy brain, the approximate left-right symmetry exists and it is used as the prior knowledge for searching the approximate pathology location. It is assumed that this area is not located symmetrically in both hemispheres, which is met in most cases. The detection algorithm was tested on 203 T2-weighted images and reached the true positive rate of 87.52% and true negative rate of 93.14%. The extraction algorithm was tested on 357 axial and 443 coronal real images from publicly available BRATS databases containing 3D volumes brain tumor patients. The results were evaluated by Dice Coefficient (axial: 0.85±0.11, coronal 0.82±0.18) and by Accuracy (axial: 0.96±0.05, coronal 0.94±0.09).

Parkinson&rsquo, s disease dysgraphia affects the majority of Parkinson&rsquo, s disease (PD) patients and is the result of handwriting abnormalities mainly caused by motor dysfunctions. Several effective approaches to quantitative PD dysgraphia analysis, such as online handwriting processing, have been utilized. In this study, we aim to deeply explore the impact of advanced online handwriting parameterization based on fractional-order derivatives (FD) on the PD dysgraphia diagnosis and its monitoring. For this purpose, we used 33 PD patients and 36 healthy controls from the PaHaW (PD handwriting database). Partial correlation analysis (Spearman&rsquo, s and Pearson&rsquo, s) was performed to investigate the relationship between the newly designed features and patients&rsquo, clinical data. Next, the discrimination power of the FD features was evaluated by a binary classification analysis. Finally, regression models were trained to explore the new features&rsquo, ability to assess the progress and severity of PD. These results were compared to a baseline, which is based on conventional online handwriting features. In comparison with the conventional parameters, the FD handwriting features correlated more significantly with the patients&rsquo, clinical characteristics and provided a more accurate assessment of PD severity (error around 12%). On the other hand, the highest classification accuracy (ACC = 97.14%) was obtained by the conventional parameters. The results of this study suggest that utilization of FD in combination with properly selected tasks (continuous and/or repetitive, such as the Archimedean spiral) could improve computerized PD severity assessment.

Neurodegenerative pathologies as Parkinson’s Disease (PD) show important distortions in speech, affecting fluency, prosody, articulation and phonation. Classically, measurements based on articulation gestures altering formant positions, as the Vocal Space Area (VSA) or the Formant Centralization Ratio (FCR) have been proposed to measure speech distortion, but these markers are based mainly on static positions of sustained vowels. The present study introduces a measurement based on the mutual information distance among probability density functions of kinematic correlates derived from formant dynamics. An absolute kinematic velocity associated to the position of the jaw and tongue articulation gestures is estimated and modeled statistically. The distribution of this feature may differentiate PD patients from normative speakers during sustained vowel emission. The study is based on a limited database of 53 male PD patients, contrasted to a very selected and stable set of eight normative speakers. In this sense, distances based on Kullback–Leibler divergence seem to be sensitive to PD articulation instability. Correlation studies show statistically relevant relationship between information contents based on articulation instability to certain motor and nonmotor clinical scores, such as freezing of gait, or sleep disorders. Remarkably, one of the statistically relevant correlations point out to the time interval passed since the first diagnostic. These results stress the need of defining scoring scales specifically designed for speech disability estimation and monitoring methodologies in degenerative diseases of neuromotor origin.

Aim: The research described is intended to give a description of articulation dynamics as a correlate of the kinematic behavior of the jaw-tongue biomechanical system, encoded as a probability distribution of an absolute joint velocity. This distribution may be used in detecting and grading speech from patients affected by neurodegenerative illnesses, as Parkinson Disease. Hypothesis: The work hypothesis is that the probability density function of the absolute joint velocity includes information on the stability of phonation when applied to sustained vowels, as well as on fluency if applied to connected speech. Methods: A dataset of sustained vowels recorded from Parkinson Disease patients is contrasted with similar recordings from normative subjects. The probability distribution of the absolute kinematic velocity of the jaw-tongue system is extracted from each utterance. A Random Least Squares Feed-Forward Network (RLSFN) has been used as a binary classifier working on the pathological and normative datasets in a leave-one-out strategy. Monte Carlo simulations have been conducted to estimate the influence of the stochastic nature of the classifier. Two datasets for each gender were tested (males and females) including 26 normative and 53 pathological subjects in the male set, and 25 normative and 38 pathological in the female set. Results: Male and female data subsets were tested in single runs, yielding equal error rates under 0.6% (Accuracy over 99.4%). Due to the stochastic nature of each experiment, Monte Carlo runs were conducted to test the reliability of the methodology. The average detection results after 200 Montecarlo runs of a 200 hyperplane hidden layer RLSFN are given in terms of Sensitivity (males: 0.9946, females: 0.9942), Specificity (males: 0.9944, females: 0.9941) and Accuracy (males: 0.9945, females: 0.9942). The area under the ROC curve is 0.9947 (males) and 0.9945 (females). The equal error rate is 0.0054 (males) and 0.0057 (females). Conclusions: The proposed methodology avails that the use of highly normalized descriptors as the probability distribution of kinematic variables of vowel articulation stability, which has some interesting properties in terms of information theory, boosts the potential of simple yet powerful classifiers in producing quite acceptable detection results in Parkinson Disease.

Up to 90 % of patients with Parkinson's disease (PD) suffer from hypokinetic dysarthria (HD). In this work, we analysed the power of conventional speech features quantifying imprecise articulation, dysprosody, speech dysfluency and speech quality deterioration extracted from a specialized poem recitation task to discriminate dysarthric and healthy speech. For this purpose, 152 speakers (53 healthy speakers, 99 PD patients) were examined. Only mildly strong correlation between speech features and clinical status of the speakers was observed. In the case of univariate classification analysis, sensitivity of 62.63% (imprecise articulation), 61.62% (dysprosody), 71.72% (speech dysfluency) and 59.60% (speech quality deterioration) was achieved. Multivariate classification analysis improved the classification performance. Sensitivity of 83.42% using only two features describing imprecise articulation and speech quality deterioration in HD was achieved. We showed the promising potential of the selected speech features and especially the use of poem recitation task to quantify and identify HD in PD.

This paper describes a preliminary investigation of Voice Pathology Detection using Deep Neural Networks (DNN). We used voice recordings of sustained vowel /a/ produced at normal pitch from German corpus Saarbruecken Voice Database (SVD). This corpus contains voice recordings and electroglottograph signals of more than 2 000 speakers. The idea behind this experiment is the use of convolutional layers in combination with recurrent Long-Short-Term-Memory (LSTM) layers on raw audio signal. Each recording was split into 64 ms Hamming windowed segments with 30 ms overlap. Our trained model achieved 71.36% accuracy with 65.04% sensitivity and 77.67% specificity on 206 validation files and 68.08% accuracy with 66.75% sensitivity and 77.89% specificity on 874 testing files. This is a promising result in favor of this approach because it is comparable to similar previously published experiment that used different methodology. Further investigation is needed to achieve the state-of-the-art results., Comment: 4 pages, 1 figure, 5 tables

Neurodegenerative pathologies produce important distortions in speech. Parkinson’s Disease (PD) leaves marks in fluency, prosody, articulation and phonation. Certain measurements based in configurations of the articulation organs inferred from formant positions, as the Vocal Space Area (VSA) or the Formant Centralization Ratio (FCR) have been classically used in this sense, but these markers represent mainly the static positions of sustained vowels on the vowel triangle. The present study proposes a measurement based on the mutual information contents of kinematic correlates derived from formant dynamics. An absolute kinematic velocity associated to the position of the articulation organs, involving the jaw and tongue is estimated and modelled statistically. The distribution of this feature is rather different in PD patients than in normative speakers when sustained vowels are considered. Therefore, articulation failures may be detected even in single sustained vowels. The study has processed a limited database of 40 female and 54 male PD patients, contrasted to a very selected and stable set of normative speakers. Distances based on Kullback-Leibler’s Divergence have shown to be sensitive to PD articulation instability. Correlation measurements show that the distance proposed shows statistically relevant relationship with certain motor and non-motor behavioral observations, as freezing of gait, or sleep disorders. These results point out to the need of defining scoring scales specifically designed for speech-based diagnose and monitoring methodologies in degenerative diseases of neuromotor origin.

International audience; This paper deals with Parkinson's disease (PD) severity estimation according to the Unified Parkinson's Disease Rating Scale: motor subscale (UPDRS III), which quantifies the hallmark symptoms of PD, using an acoustic analysis of speech signals. Experimental dataset comprised 42 speech tasks acquired from 50 PD patients (UPDRS III ranged from 6 to 92). It was divided into subsets: words, sentences, reading text, monologue and diadochokinetic tasks. We performed a parametrization of the whole corpus and these groups separately using a wide range of conventional and novel speech features. We used guided regu-larized random forest algorithm to select features with maximum clinical information and performed random forests regression to estimate PD severity. According to significant correlations between true UPDRS III scores and scores predicted by the proposed methodology it was shown that information extracted through variety of speech tasks can be used to estimate the degree of PD severity.

Background: Impairment of speech prosody is characteristic for Parkinson's disease (PD) and does not respond well to dopaminergic treatment. Objectives: We assessed whether baseline acoustic parameters, alone or in combination with other predominantly non-dopaminergic symptoms may predict global cognitive decline as measured by the Addenbrooke's cognitive examination (ACE-R) and/or worsening of cognitive status as assessed by a detailed neuropsychological examination. Methods: Forty-four consecutive non-depressed PD patients underwent clinical and cognitive testing, and acoustic voice analysis at baseline and at the two-year follow-up. Influence of speech and other clinical parameters on worsening of the ACE-R and of the cognitive status was analyzed using linear and logistic regression. Results: The cognitive status (classified as normal cognition, mild cognitive impairment and dementia) deteriorated in 25% of patients during the follow-up. The multivariate linear regression model consisted of the variation in range of the fundamental voice frequency (F0VR) and the REM Sleep Behavioral Disorder Screening Questionnaire (RBDSQ). These parameters explained 37.2% of the variability of the change in ACE-R. The most significant predictors in the univariate logistic regression were the speech index of rhythmicity (SPIR; p = 0.012), disease duration (p = 0.019), and the RBDSQ (p = 0.032). The multivariate regression analysis revealed that SPIR alone led to 73.2% accuracy in predicting a change in cognitive status. Combining SPIR with RBDSQ improved the prediction accuracy of SPIR alone by 73%. Conclusions: Impairment of speech prosody together with symptoms of RBD predicted rapid cognitive decline and worsening of PD cognitive status during a two-year period. (C) 2016 Elsevier Ltd. All rights reserved.

The paper focuses on the segmentation of magnetic resonance imaging (MRI) slices and 3D modelling of the temporomandibular joint disc in order to help physicians diagnose patients with dysfunction of the temporomandibular joint (TMJ). The TMJ is one of the most complex joints in the human body. The most common joint dysfunction is due to the disc. The disc is a soft tissue, which in principle cannot be diagnosed by the CT method. Therefore, a 3D model is made from the MRI slices, which can image soft tissues. For the segmentation of the disc in individual slices a new method is developed based on spatial distribution and anatomical TMJ structure with automatic thresholding. The thresholding is controlled by a genetic algorithm. The 3D model is realized using the marching cube method.

The article deals with a method for segmenting the articular condyle of the temporomandibular joint from magnetic resonance images. Temporomandibular condyle is segmented by the modified method of marker-controlled watershed. This method is improved by an algorithm for the creation of labels and boundaries of objects based on the analysis of image data. Creating a 3D model has been improved by an algorithm for oversampling the object in the z-axis by cubic spline.

This paper deals with an acoustic analysis of hypokinetic dysarthria in patients with Parkinson's disease (PD). The analysis is based on parametrization of five basic Czech vowels using conventional features and parameters based on empirical mode decomposition (EMD). Experimental dataset consists of 84 PD patients with different disease progress and 49 healthy controls. From the single-vowel-analysis point of view we observed that sustained vowels pronounced with minimum intensity (not whispering) outperformed the other vowels' realization (including the most popular sustained vowel [a] pronounced with normal intensity). Then we employed a classification along with feature selection and again obtained the best results in the case of silent sustained vowels (accuracy ACC = 84 %, sensitivity SEN = 86% and specificity SPE = 82 %). Finally we considered classification of PD using different vowels' realization and reached accuracy = 94 %, sensitivity = 96% and specificity = 90 %. Features based on EMD significantly improved the results.

HighlightsWe analysed neutral, stress-modified and rhymed speech in Parkinson's disease.We proposed quantitative prosodic analysis of poem recitation task.We showed rhythmical demands improve identification of hypokinetic dysarthria.We introduced a concept of permutation test in dysarthric speech analysis. Background and objectiveHypokinetic dysarthria (HD) is a frequent speech disorder associated with idiopathic Parkinson's disease (PD). It affects all dimensions of speech production. One of the most common features of HD is dysprosody that is characterized by alterations of rhythm and speech rate, flat speech melody, and impairment of speech intensity control. Dysprosody has a detrimental impact on speech naturalness and intelligibility. MethodsThis paper deals with quantitative prosodic analysis of neutral, stress-modified and rhymed speech in patients with PD. The analysis of prosody is based on quantification of monopitch, monoloudness, and speech rate abnormalities. Experimental dataset consists of 98 patients with PD and 51 healthy speakers. For the purpose of HD identification, sequential floating feature selection algorithm and random forests classifier is used. In this paper, we also introduce a concept of permutation test applied in the field of acoustic analysis of dysarthric speech. ResultsProsodic features obtained from stress-modified reading task provided higher classification accuracies compared to the ones extracted from reading task with neutral emotion demonstrating the importance of stress in speech prosody. Features calculated from poem recitation task outperformed both reading tasks in the case of gender-undifferentiated analysis showing that rhythmical demands can in general lead to more precise identification of HD. Additionally, some gender-related patterns of dysprosody has been observed. ConclusionsThis paper confirms reduced variation of fundamental frequency in PD patients with HD. Interestingly, increased variability of speech intensity compared to healthy speakers has been detected. Regarding speech rate disturbances, our results does not report any particular pattern. We conclude further development of prosodic features quantifying the relationship between monopitch, monoloudness and speech rate disruptions in HD can have a great potential in future PD analysis.

The feature selection is inevitable part of machine learning techniques in biomedical engineering and bioinformatics. Feature selection methods are used to select the most discriminative features, e.g. for disease classification. Even if there are plenty of feature selection methods the stability of these algorithms is still open question. Another issue with assessing the stability of feature selection is that there are several stability measures providing different views on stability. Here, we compare well-known stability measures and evaluate their performance on artificial and real data.

The paper presents novel approach to automatic identification of successful phone calls in call centers by means of dialogue features. This approach can be very useful to spot the unsuccessful sessions within the large database of recorded telephone calls and can help the supervisors of call centers to figure out mistakes made by their agents. The features used for decision making are extracted from four cues namely hesitation, reaction, Interruption and cumulative voice activity. The results achieved suggested that these features have a strong discriminative power in terms of classification between successful and unsuccessful phone calls showing F-measure of 96% by using Naive Bayesian Classifier.

Feature selection techniques become significant part of many bioinformatics and biomedical applications. Choosing the important features is essential for biomarker discovery, provide better understanding of the data and potentially improve prediction performance. However, as the number of samples in dataset is small, the feature selection tends to be unstable. In this paper, the stability of five popular feature selection techniques is compared and analyzed when original dataset is perturbed by adding, removing or simply resampling the original observations. Next, the feature selection techniques are used as filter prior to AdaBoost classifier and their influence on classification accuracy and Mathews correlation coefficient is compared.

Parkinson’s disease (PD) is a neurodegenerative disorder which impairs motor skills, speech, and other functions such as behavior, mood, and cognitive processes. One of the most typical clinical hallmarks of PD is handwriting deterioration, usually the first manifestation of PD. The aim of this study is twofold: (a) to find a subset of handwriting features suitable for identifying subjects with PD and (b) to build a predictive model to efficiently diagnose PD. We collected handwriting samples from 37 medicated PD patients and 38 age- and sex-matched controls. The handwriting samples were collected during seven tasks such as writing a syllable, word, or sentence. Every sample was used to extract the handwriting measures. In addition to conventional kinematic and spatio-temporal handwriting measures, we also computed novel handwriting measures based on entropy, signal energy, and empirical mode decomposition of the handwriting signals. The selected features were fed to the support vector machine classifier with radial Gaussian kernel for automated diagnosis. The accuracy of the classification of PD was as high as 88.13%, with the highest values of sensitivity and specificity equal to 89.47% and 91.89%, respectively. Handwriting may be a valuable marker as a diagnostic and screening tool.

HighlightsWe exploit the fact that movement during handwriting of a text consists of on-surface and in-air movements and use it for differential diagnosis of PD.By applying feature selection algorithms and support vector machine learning methods to separate PD from HC, we demonstrated 85% classification accuracy.Decision support system based on handwriting analysis can be complementary method to diagnosis made by clinician or other decision support systems. Background and objectiveParkinson's disease (PD) is the second most common neurodegenerative disease affecting significant portion of elderly population. One of the most frequent hallmarks and usually also the first manifestation of PD is deterioration of handwriting characterized by micrographia and changes in kinematics of handwriting. There is no objective quantitative method of clinical diagnosis of PD. It is thought that PD can only be definitively diagnosed at postmortem, which further highlights the complexities of diagnosis. MethodsWe exploit the fact that movement during handwriting of a text consists not only from the on-surface movements of the hand, but also from the in-air trajectories performed when the hand moves in the air from one stroke to the next. We used a digitizing tablet to assess both in-air and on-surface kinematic variables during handwriting of a sentence in 37 PD patients on medication and 38 age- and gender-matched healthy controls. ResultsBy applying feature selection algorithms and support vector machine learning methods to separate PD patients from healthy controls, we demonstrated that assessing the in-air/on-surface hand movements led to accurate classifications in 84% and 78% of subjects, respectively. Combining both modalities improved the accuracy by another 1% over the evaluation of in-air features alone and provided medically relevant diagnosis with 85.61% prediction accuracy. ConclusionsAssessment of in-air movements during handwriting has a major impact on disease classification accuracy. This study confirms that handwriting can be used as a marker for PD and can be with advance used in decision support systems for differential diagnosis of PD.

One of the most frequent clinical hallmarks of Parkinson's disease (PD) is micrographia. Micrographia in PD is characterized by the decreased letter size and by changes in the kinematic aspects including increased movement time, decreased velocities and accelerations, and increased number of changes in velocity and acceleration. Based on the literature survey we proposed template to acquire handwriting during different tasks. In addition to well established tasks for PD diagnosis such as Archimedean spiral, we designed new tasks to acquire all aspects of micrographia. The database consists of eight different handwriting samples from seventy-five subjects. The presented results shows almost 80% overall classification accuracy.

Parkinsons disease (PD) is neurodegenerative disorder with very high prevalence rate occurring mainly among elderly. One of the most typical symptoms of PD is deterioration of handwriting that is usually the first manifestation of Parkinsons disease. In this study, a new modality - in-air trajectory during handwriting - is proposed to efficiently diagnose PD. Experimental results showed that analysis of in-air trajectories is capable of assessing subtle motor abnormalities that are connected with PD. Moreover, conjunction of in-air trajectories with conventional on-surface handwriting allows us to build predictive model with PD classification accuracy over 80%. In total, we compute over 600 handwriting features. Then, we select smaller subset of these features using two feature selection algorithms: Mann-Whitney U-test filter and relief algorithm, and map these feature subsets to binary classification response using support vector machines.

This article deals with methods for automatic detection of moving objects in non-standard situations in secure areas, such as nuclear power plants, storage of hazardous materials, etc. Speed and direction, respectively, slow motion to a stop is detected for the monitored object. Specifically: the automatic detection of movement in the opposite direction than normal, passage or move too fast, and stopping or blocking passages. The emphasis is put on real-time processing and on design effective and efficient methods. A new approach is a combination and modification of optical flow techniques and Mixture Of Gaussians method (MOG).

The article is focused on a new method for automatic detection of significant curves of face in areas of the upper and lower jaws. Detecting significant points and facial lines is important for preoperative examination and preparation of the patient having an operation in the field of orthognathic surgery. Oral and maxillofacial surgery is a process in which congenital dentofacial deformities, posttraumatic and post resection deformities are corrected with surgical operations of the facial skeleton, sometimes combined with various soft tissue procedures. A new method for the automated detection of facial lines in X-ray images will be used to diagnose the disease and to accelerate and facilitate the planning of a specific individual surgical procedure.