Search

Pseudomonas (P.) aeruginosa is a bacterium responsible for severe infections that have become a real concern in hospital environments. Nosocomial infections caused by P. aeruginosa are often hard to treat because of its intrinsic resistance and remarkable ability to acquire further resistance mechanisms to multiple groups of antimicrobial agents. Thus, rapid determination of the susceptibility of P. aeruginosa isolates to antibiotics is crucial for effective treatment. The current methods used for susceptibility determination are time-consuming; hence the importance of developing a new method. Fourier-transform infra-red (FTIR) spectroscopy is known as a rapid and sensitive diagnostic tool, with the ability to detect minor abnormal molecular changes including those associated with the development of antibiotic- resistant bacteria. The main goal of this study is to evaluate the potential of FTIR spectroscopy together with machine learning algorithms, to determine the susceptibility of P. aeruginosa to different antibiotics in a time span of ∼20 min after the first culture. For this goal, 590 isolates of P. aeruginosa, obtained from different infection sites of various patients, were measured by FTIR spectroscopy and analyzed by machine learning algorithms. We have successfully determined the susceptibility of P. aeruginosa to various antibiotics with an accuracy of 82-90%.

Significance: Accurate and objective identification of Alzheimer’s disease (AD) and dementia with Lewy bodies (DLB) is of major clinical importance due to the current lack of low-cost and noninvasive diagnostic tools to differentiate between the two. Developing an approach for such identification can have a great impact in the field of dementia diseases as it would offer physicians a routine objective test to support their diagnoses. The problem is especially acute because these two dementias have some common symptoms and characteristics, which can lead to misdiagnosis of DLB as AD and vice versa, mainly at their early stages. Aim: The aim is to evaluate the potential of mid-infrared (IR) spectroscopy in tandem with machine learning algorithms as a sensitive method to detect minor changes in the biochemical structures that accompany the development of AD and DLB based on a simple peripheral blood test, thus improving the diagnostic accuracy of differentiation between DLB and AD. Approach: IR microspectroscopy was used to examine white blood cells and plasma isolated from 56 individuals: 26 controls, 20 AD patients, and 10 DLB patients. The measured spectra were analyzed via machine learning. Results: Our encouraging results show that it is possible to differentiate between dementia (AD and DLB) and controls with an ∼86% success rate and between DLB and AD patients with a success rate of better than 93%. Conclusions: The success of this method makes it possible to suggest a new, simple, and powerful tool for the mental health professional, with the potential to improve the reliability and objectivity of diagnoses of both AD and DLB.

Human viral and bacterial infections are responsible for a variety of diseases that are still the main causes of death and economic burden for society across the globe. Despite the different responses of the immune system to these infections, some of them have similar symptoms, such as fever, sneezing, inflammation, vomiting, diarrhea, and fatigue. Thus, physicians usually encounter difficulties in distinguishing between viral and bacterial infections on the basis of these symptoms. Rapid identification of the etiology of infection is highly important for effective treatment and can save lives in some cases. The current methods used for the identification of the nature of the infection are mainly based on growing the infective agent in culture, which is a time-consuming (over 24 h) and usually expensive process. The main objective of this study was to evaluate the potential of the mid-infrared spectroscopic method for rapid and reliable identification of bacterial and viral infections based on simple peripheral blood samples. For this purpose, white blood cells (WBCs) and plasma were isolated from the peripheral blood samples of patients with confirmed viral or bacterial infections. The obtained spectra were analyzed by multivariate analysis: principle component analysis (PCA) followed by linear discriminant analysis (LDA), to identify the infectious agent type as bacterial or viral in a time span of about 1 h after the collection of the blood sample. Our preliminary results showed that it is possible to determine the infectious agent with high success rates of 82% for sensitivity and 80% for specificity, based on the WBC data.

Rapid identification of bacterial infection is very important and in many cases can save human life. Many pathogens can cause infections. While these infections share identical symptoms, the immune system responds differently to these pathogens. The current microbiology lab methods used to diagnose the infection type are time consuming (2-4 days). Thus, physicians may be tempted to start unnecessary antibiotic treatment, based on their wrong diagnosis (based on experience) of the infection. Uncontrolled use of antibiotics is the main driving force for the development of multi drug resistant bacteria which is considered a global health problem. We hypothesize that the different responses of the immune system to the infecting pathogens, cause some minute biochemical changes in the blood componentsthat can be detected by infrared spectroscopy which is known as a fast, accurate, sensitive and low cost method. In this study, we used infrared microscopy to measure the vibrational spectra of white blood cells (WBC) samples of 105 infected patients (69 bacterial and 36 with viral infection) and 90 controls (non-infected patients). The obtained spectra were analyzed using machine learning algorithms to identify the infection type as bacterial or viral in a time span of less than one hour after blood sample collection. Our study results showed that it is possible to determine the infection type with high success rates of 93% sensitivity and 85% specificity, based solely on WBC obtained from simple peripheral blood samples.

The spread of multidrug resistant bacteria has become a global concern. One of the most important and emergent classes of multidrug-resistant bacteria is extended-spectrum β-lactamase-producing bacteria (ESBL-positive = ESBL+). Due to widespread and continuous evolution of ESBL-producing bacteria, they become increasingly resistant to many of the commonly used antibiotics, leading to an increase in the mortality associated with resulting infections. Timely detection of ESBL-producing bacteria and rapid determination of their susceptibility to appropriate antibiotics can reduce the spread of these bacteria and the consequent complications. Routine methods used for the detection of ESBL-producing bacteria are time-consuming, requiring at least 48 h to obtain results. In this study, we evaluated the potential of infrared spectroscopic microscopy, combined with multivariate analysis for rapid detection of ESBL-producing Escherichia coli ( E. coli) isolated from urinary-tract infection (UTI) samples. Our measurements were conducted on 837 samples of uropathogenic E. coli (UPEC), including 268 ESBL+ and 569 ESBL-negative (ESBL-) samples. All samples were obtained from bacterial colonies after 24 h culture (first culture) from midstream patients' urine. Our results revealed that it is possible to detect ESBL-producing bacteria, with a 97% success rate, 99% sensitivity, and 94% specificity for the tested samples, in a time span of few minutes following the first culture.

Correction for ‘Diagnosis of inaccessible infections using infrared microscopy of white blood cells and machine learning algorithms’ by Adam H. Agbaria et al., Analyst, 2020, DOI: 10.1039/D0AN00752H.

This paper focuses on estimating the quality of a clustering process. In our case - the task is to cluster short speech segments that belong to different speakers. Moreover, speaker clustering quality may be well estimated on several clustering approaches if they all based on the same features. This is very important, as it allows us to use the same quality estimation system without retraining, and achieve reasonable results even when the clustering method is changed. We predict the system’s quality by applying a logistic regression estimator on a several statistical parameters of the clustering. In this paper, mean-shift clustering with either cosine or probabilistic linear discriminant analysis (PLDA) score as similarity measure, and stochastic vector quantization (VQ) with cosine distance were applied in order to cluster the short speaker segments represented by i-vectors. The quality of the clustering is measured using the average cluster purity (ACP), average speaker purity (ASP) and K. We show that these measures can be estimated fairly well by applying logistic regression based on various clustering statistics that calculated once clustering is over. These statistical parameters are used as a feature vector representing the clustering.

Short segments speaker clustering has significant importance both for diarization and applications such as short push-to-tatk (PTT) segments clustering. In this paper we present a new way to cluster speech segments by applying a stochastic vector quantization (VQ) with a cosine metric together with a speaker clustering quality estimator based on logistic regression. The VQ is performed on codebooks of different sizes, and the choice of the best clustering result is estimated using logistic regression. The algorithm is tested on a large range of speakers, between 2 to 60. The results are compared to those of the mean-shift clustering method, which was already tested for this task several times. The results are a bit below those of the cosine similarity measure-based mean-shift clustering. The advantage is in the run-time which is approximately 10 times faster.

Early detection of soil-borne pathogens, which have a negative effect on almost all agricultural crops, is crucial for effective targeting with the most suitable antifungal agents and thus preventing and/or reducing their severity. They are responsible for severe diseases in various plants, leading in many cases to substantial economic losses. In this study, infrared (IR) spectroscopic method, which is known as sensitive, accurate and rapid, was used to discriminate between different fungi in a mixture was evaluated. Mixed and pure samples of Colletotrichum, Verticillium, Rhizoctonia, and Fusarium genera were measured using IR microscopy. Our spectral results showed that the best differentiation between pure and mixed fungi was obtained in the 675–1800 cm−1 wavenumber region. Principal components analysis (PCA), followed by linear discriminant analysis (LDA) as a linear classifier, was performed on the spectra of the measured classes. Our results showed that it is possible to differentiate between mixed-calculated categories of phytopathogens with high success rates (~100%) when the mixing percentage range is narrow (40–60) in the genus level; when the mixing percentage range is wide (10–90), the success rate exceeded 85%. Also, in the measured mixed categories of phytopathogens it is possible to differentiate between the different categories with ~100% success rate.

This paper extends upon previous work using Mean Shift algorithm to perform speaker clustering on i-vectors generated from short speech segments. In this paper we examine the effectiveness of Spherical Normalization in the presence of different numbers of speakers. This normalization method is not only easy to implement but also improves clustering results. The main improvement is that the proposed mean shift algorithm is more robust to changes in the number of speakers. In the case of 30 speakers, we achieved evaluation parameter K of 75.3 compared to 72.1 with the baseline system.

Fungi are microorganisms that are divided into groups and subgroups according to their similarity, genera, species, and strains. Fusarium is considered a phytopathogen that attacks variety of crops throughout the world, causing diseases resulting in severe economic losses. Many of the Fusarium species cause similar symptoms, making it impossible to distinguish among them based on symptoms alone. Fungicides are commonly the most effective treatment of these pathogens, and their use is effective and could prevent or decrease its severity and spread when detected early. Currently, classical methods (microbiological, molecular) are time-consuming. In this study, we aimed to distinguish among three different groups: Fusarium oxysporum, Fusarium solani, and a mixture of both species. Thus, we used Fourier transform infrared microscopy combined with principal component analysis (PCA) and linear discriminant analysis (LDA) classifier. Using the first ten PCs, our classification results showed a 94% success rate in distinguishing among the three groups.

Fungi are considered as serious pathogens for many plants, potentially causing severe economic damage. Early detection and identification of these pathogens is crucial for their timely control. The methods available for identification of fungi are time consuming and not always very specific. In this study, the potential of FTIR-ATR spectroscopy was examined together with advanced mathematical principle component analysis (PCA) and statistical linear discriminant analysis (LDA) to differentiate among 10 isolates of Fusarium oxysporum. The results are encouraging and indicate that FTIR-ATR can successfully detect different isolates of Fusarium oxysporum. Based on PCA and LDA calculations in the region 850–1775 cm-1with 16 PC's, the different strains from the same fungal genus could be classified with 75.3% and 69.5% success rates using the “leave one out” method and “20–80% algorithm” respectively.

International audience

Fourier Transform Infrared (FTIR) and Raman spectroscopies have emerged as powerful tools for chemical analysis. This is due to their ability to provide detailed information about the spatial distribution of chemical composition at the molecular level. A biological sample, i.e. bacteria or fungi, has a typical spectrum. This spectral fingerprint, characterizes the sample and can therefore be used for differentiating between biology samples which belong to different groups, i.e., several different isolates of a given fungi. When the spectral differences between the groups are minute, multivariate analysis should be used to provide a good differentiation. We hereby review several results which demonstrate the differentiation success obtained by combining spectroscopy measurements and multivariate analysis.

In forensic voice comparison, it is strongly recommended to follow the Bayesian paradigm to present a forensic evidence to the court. In this paradigm, the strength of the forensic evidence is summarized by a likelihood ratio (LR). But in the real world, to base only on the LR without looking to its degree of reliability does not allow experts to have a good judgement. This work is mainly motivated by the need to quantify this reliability. In this concept, we think that the presence of speaker specific information and its homogeneity between the two signals to compare should be evaluated. This paper is dedicated to the latter, the homogeneity. We propose an information theory based homogeneity measure which determines whether a voice comparison is feasible or not.

The primary objective of speaker diarization system is to designate speech segments to one of K speakers in the conversation. We use a hidden-distortion-model (HDM)-based system. HDM allows using different emission models as speaker models. We investigate the effect of randomization in two different levels. One level is stochastic training versus deterministic training and the other, random model initialization versus preserving initialization from the previous iteration. The emission models were codebooks (CBs) trained using K-means algorithm, both, batch and stochastic versions, as well as a self-organizing map (SOM) in its stochastic version. The evaluation performed on 108 telephone conversations from the LDC CallHome corpus. We will show that randomizing is always outperforming the deterministic training. Stochastic training demonstrated relative improvement of 3.5%. Random initialization achieved relative improvement of 7.28% comparing to preservation of initialization from the previous iteration.

In the present study we examined the potential of Fourier Transformed Infrared (FTIR) spectroscopy for accurate identification and differentiation of mixed bacteria samples in a time span of a few minutes. The bacterial samples used in this study are Escherichia (E.) coli, Bacillus (B.) megaterium and a mixture of E. coli and B. megaterium. The best results of differentiation were obtained within the 675–1800 cm−1 range. In this range the dimension of the feature vector is 293. Principal components analysis (PCA) followed by linear discriminant analysis (LDA) as a linear classifier were performed on the spectra of the three measured classes. When differentiating between the pure sets of E. coli and B. megaterium, 100% success was obtained for a feature vector composed of the first 12 principal components (PCs). An error rate of less than 2% was achieved taking only the first 20 PCs among the three categories of samples.

The primary objective of any speaker diarization system is to designate speech segments to one of K speakers in the conversation. In this work we will focus on telephone conversations, where the number of speakers is given and equal 2. We use a hidden-distortion-model (HDM)-based system. HDM allows using different emission models as speaker models. The choice of adequate emission models, properly representing the data characteristics is important for the systems' performance. We investigate the effect of several codebooks (CBs) based emission models, with Euclidian and Mahalanobis distances. The Mahalanobis distance was chosen due its potential to produce a better representation of the data's spatial layout, while limitations where maid to retain the model from divergence. The influence of the different methods is evaluated using 108 telephone conversations taken from the LDC CallHome corpus. All the experiments achieved results poorer than the original SOM-based system (DER=12.70%).

In this work we examine whether linear discriminant analysis (LDA) can improve the diarization performance, when used as an additional phase in a telephone conversation diarization system. We first apply a classical diarization system. Using systems output (to define the classes of interest) an LDA transformation on the mel-cepstrum features is performed. Then, the final diarization process is applied onto the transformed features. A relative improvement of 14.8% was obtained on LDC America CallHome database. The LDA seemed sensible to both segment duration and amount of data available for training, as shown by the results obtained on NIST SRE-05 database where no significative improvement was observed.

We examine different initializations and their influence on the performances of iterative speaker diarization system. Six methods of initializations were under examination, starting with a naive frame based random initialization, continue with uniform conversation dividing between the clusters and ending with weighted segmental k-means. The initialization methods were tested on two telephone conversation databases: LDC America CallHome and NIST SRE-05. In contrast to most works on meeting and shows where the speakers turns are not very frequent and minimal duration constraints of 2.5 sec or more can be applied to capture speakers statistics, in telephone conversations the speaker turns are much more frequent and the minimum duration should be set to several hundreds of milliseconds. In such cases, good cluster initialization is very important. It will be shown that good initialization using weighted segmental k-means is outperforms all other methods, and the either fixed or minimum duration constraints can be minor, and even without any constraint on the segment duration the results are significantly better than in other initializations.

Colletotrichum coccodes (C. coccodes) is a pathogenic fungus which causes anthracnose on tomatoes and black dot disease in potatoes. It is important to differentiate among these isolates and to detect the origin of newly discovered isolates, in order to treat the disease in its early stages. However, distinguishing between isolates using common biological methods is time-consuming, and not always available. We used Fourier Transform Infra-Red (FTIR)-Attenuated Total Reflectance (ATR) spectroscopy and advanced mathematical and statistical methods to distinguish between different isolates of C. coccodes. To our knowledge, this is the first time that FTIR-ATR spectroscopy was used, combined with multivariate analysis, to classify such a large number of 15 isolates belonging to the same species. We obtained a success rate of approximately 90% which was achieved using the region 800-1775 cm(-1). In addition we succeeded in determining the relative spectral similarity between different fungal isolates by developing a new algorithm. This method could be an important potential diagnostic tool in agricultural research, since it may outline the extent of the biological similarity between fungal isolates. Based on the PCA calculations, we grouped the fifteen isolates included in this study into four different degrees of similarity.

The early diagnosis of phytopathogens is of a great importance; it could save large economical losses due to crops damaged by fungal diseases, and prevent unnecessary soil fumigation or the use of fungicides and bactericides and thus prevent considerable environmental pollution. In this study, 18 isolates of three different fungi genera were investigated; six isolates of Colletotrichum coccodes, six isolates of Verticillium dahliae and six isolates of Fusarium oxysporum. Our main goal was to differentiate these fungi samples on the level of isolates, based on their infrared absorption spectra obtained using the Fourier transform infrared-attenuated total reflection (FTIR-ATR) sampling technique. Advanced statistical and mathematical methods: principal component analysis (PCA), linear discriminant analysis (LDA), and k-means were applied to the spectra after manipulation. Our results showed significant spectral differences between the various fungi genera examined. The use of k-means enabled classification between the genera with a 94.5% accuracy, whereas the use of PCA [3 principal components (PCs)] and LDA has achieved a 99.7% success rate. However, on the level of isolates, the best differentiation results were obtained using PCA (9 PCs) and LDA for the lower wavenumber region (800-1775 cm(-1)), with identification success rates of 87%, 85.5%, and 94.5% for Colletotrichum, Fusarium, and Verticillium strains, respectively.

Fusarium is a large fungi genus of a large variety of species and strains which inhabits soil and vegetation. It is distributed worldwide and affiliated to both warm and cold weather. Fusarium oxysporum species, for instance, cause the Fusarium wilt disease of plants, which appears as a leaf wilting, yellowing and eventually plant death. Early detection and identification of these pathogens are very important and might be critical for their control. Previously, we have managed to differentiate among different fungi genera (Rhizoctonia, Colletotrichum, Verticillium and Fusarium) using FTIR-ATR spectroscopy methods and cluster analysis. In this study, we used Fourier-transform infrared (FTIR) attenuated total reflection (ATR) spectroscopy to discriminate and differentiate between different strains of F. oxysporum. The result obtained was of spectral patterns distinct to each of the various examined strains, which belong to the same species. These differences were not as significant as those found between the different genera species. We applied advanced statistical techniques: principal component analysis (PCA) and linear discriminant analysis (LDA) on the FTIR-ATR spectra in order to examine the feasibility of distinction between these fungi strains. The results are encouraging and indicate that the FTIR-ATR methodology can differentiate between the different examined strains of F. oxysporum with a high success rate. Based on our PCA and LDA calculations performed in the regions [900–1775 cm−1, 2800–2990 cm−1, with 9 PCs], we were able to classify the different strains with high success rates: Foxy1 90%, Foxy2 100%, Foxy3 100%, Foxy4 92.3%, Foxy5 83.3% and Foxy6 100%.

In this paper we employ backward Viterbi search for speech recognition. Contrary to forward Viterbi search that is performed from the beginning to the end, and where a word depends on the preceding words, backward Viterbi search is performed from the end to the beginning and the current word depends from the following words. As the errors of the forward and the backward searches are not the same, improvement can be achieved by combining the forward and the backward Viterbi search. The fusion is attained by an expert system based on rover algorithm, and using confidence measure for the words and optimal confidence value for null arcs depending on its place in word transition network (WTN). The experimental result of the combined system showed significant improvement over both forward and backward Viterbi decoding system on the Number 95 database.

This paper presents a comparison between several features normalization methods, and a comparison between different types of Gaussian Mixture Model (GMM) based supervectors normalizations for robust Speaker Verification. We implemented the methods of normalizations as a part of speaker verification system using Support Vector Machine (SVM) classifier and GMM-based supervectors. When implementing the speaker recognition system, we used Mel Frequency Cepstral Coefficients (MFCC) feature extraction. A valid question is which features normalization to use, if any. We examine the most common methods of feature normalizations, such as: Feature Warping mapping, and Cepstral Mean Subtraction (CMS) normalization with and without variance normalization. These methods were compared to features without normalization at all, and to a basic [-1, 1] normalization. In addition, we applied few types of normalizations to the GMM-mean supervectors, in order to improve the performance of the SVM classifier. All comparisons of the speaker verification system had been done in terms of DET curve, EER (Equal Error Rate) and Min. DCF. The best results we achieved were on combined supervector normalizations of Universal Background Model (UBM) Standard Deviation (STD) and [-1, 1] normalization. The type of the MFCC normalization has no big influence on the verification performance. The best results were: EER about 5.0% and MIN. DCF of 0.02.

One inherent deficiency of most diarization systems is their inability to handle co-channel or overlapped speech. Most of the suggested algorithms perform under singular conditions, require high computational complexity in both time and frequency domains. In this study, frame based entropy analysis of the audio data in the time domain serves as a single feature for an overlapped speech detection algorithm. Identification of overlapped speech segments is performed using Gaussian Mixture Modeling (GMM) along with well known classification algorithms applied on two speaker conversations. By employing this methodology, the proposed method eliminates the need for setting a hard threshold for each conversation or database. LDC CALLHOME American English corpus is used for evaluation of the suggested algorithm. The proposed method successfully detects 63.2% of the frames labeled as overlapped speech by the manual segmentation, while keeping a 5.4% false-alarm rate.

Speaker diarization systems attempt to assign temporal speech segments in a conversation to the appropriate speaker, and non-speech segments to non-speech. Speaker diarization systems basically provide an answer to the question “Who spoke when ?”.

In this paper a piecewise-dependent-data (PDD) clustering algorithm is presented, and a proof of its convergence to a local minimum is given. A distortion measure-based model represents each cluster. The proposed algorithm is iterative. At the end of each iteration, a competition between the models is performed. Then the data is regrouped between the models. The “movement” of the data between the models and the retraining allows the minimization of the overall system distortion. The Kohonen Self-Organizing Map (SOM) was used as the VQ model for clustering. The clustering algorithm was tested using data generated from four generators of Continuous Density HMM (CDHMM). It was demonstrated that the overall distortion is a decreasing function.

In many time-series such as speech, biosignals, protein chains, etc. there is a dependency between consecutive vectors. As the dependency is limited in duration, such data can be referred to as piecewise-dependent data (PDD). In clustering, it is frequently needed to minimize a given distance function. In this letter, we will show that in PDD clustering there is a contradiction between the desire for high resolution (short segments and low distance) and high accuracy (long segments and high distance), i.e., meaningful clustering.

Speaker diarization systems attempt segmentation and labeling of a conversation between R speakers, while no prior information is given regarding the conversation. Most state of the art diarization systems require the full body of the conversation data prior to the application of some diarization approach. However, for some applications such as forensics, which handles vast amount of data, an on-line or incremental diarization is of high importance. For that purpose, a two-stage incremental diarization of telephone conversations algorithm is suggested. On the first stage, a fully unsupervised diarization algorithm is applied over an initial training segment from the conversation. The secondstage is composed of time-series clustering of increments of the conversation. Applying incremental diarization over 1802 telephone conversations from NIST 2005 SER generated an increase in diarization error of approximately 2% compared to the diarization error of an off-line diarization system.