Classifiers for Ischemic Stroke Lesion Segmentation: A Comparison Study

Motivation

Ischemic stroke, triggered by an obstruction in the cerebral blood supply, leads to infarction of the affected brain tissue. An accurate and reproducible automatic segmentation is of high interest, since the lesion volume is an important end-point for clinical trials. However, various factors, such as the high variance in lesion shape, location and appearance, render it a difficult task.

Methods

In this article, nine classification methods (e.g. Generalized Linear Models, Random Decision Forests and Convolutional Neural Networks) are evaluated and compared with each other using 37 multiparametric MRI datasets of ischemic stroke patients in the sub-acute phase in terms of their accuracy and reliability for ischemic stroke lesion segmentation. Within this context, a multi-spectral classification approach is compared against mono-spectral classification performance using only FLAIR MRI datasets and two sets of expert segmentations are used for inter-observer agreement evaluation.

Results and Conclusion

The results of this study reveal that high-level machine learning methods lead to significantly better segmentation results compared to the rather simple classification methods, pointing towards a difficult non-linear problem. The overall best segmentation results were achieved by a Random Decision Forest and a Convolutional Neural Networks classification approach, even outperforming all previously published results. However, none of the methods tested in this work are capable of achieving results in the range of the human observer agreement and the automatic ischemic stroke lesion segmentation remains a complicated problem that needs to be explored in more detail to improve the segmentation results.     

A Generalizable Brain-Computer Interface (BCI) Using Machine Learning for Feature Discovery

This work describes a generalized method for classifying motor-related neural signals for a brain-computer interface (BCI), based on a stochastic machine learning method. The method differs from the various feature extraction and selection techniques employed in many other BCI systems. The classifier does not use extensive a-priori information, resulting in reduced reliance on highly specific domain knowledge. Instead of pre-defining features, the time-domain signal is input to a population of multi-layer perceptrons (MLPs) in order to perform a stochastic search for the best structure. The results showed that the average performance of the new algorithm outperformed other published methods using the Berlin BCI IV (2008) competition dataset and was comparable to the best results in the Berlin BCI II (2002–3) competition dataset. The new method was also applied to electroencephalography (EEG) data recorded from five subjects undertaking a hand squeeze task and demonstrated high levels of accuracy with a mean classification accuracy of 78.9% after five-fold cross-validation. Our new approach has been shown to give accurate results across different motor tasks and signal types as well as between subjects.     

A Rapid Method for Characterization of Protein Relatedness Using Feature Vectors

We propose a feature vector approach to characterize the variation in large data sets of biological sequences. Each candidate sequence produces a single feature vector constructed with the number and location of amino acids or nucleic acids in the sequence. The feature vector characterizes the distance between the actual sequence and a model of a theoretical sequence based on the binomial and uniform distributions. This method is distinctive in that it does not rely on sequence alignment for determining protein relatedness, allowing the user to visualize the relationships within a set of proteins without making a priori assumptions about those proteins. We apply our method to two large families of proteins: protein kinase C, and globins, including hemoglobins and myoglobins. We interpret the high-dimensional feature vectors using principal components analysis and agglomerative hierarchical clustering. We find that the feature vector retains much of the information about the original sequence. By using principal component analysis to extract information from collections of feature vectors, we are able to quickly identify the nature of variation in a collection of proteins. Where collections are phylogenetically or functionally related, this is easily detected. Hierarchical agglomerative clustering provides a means of constructing cladograms from the feature vector output.     

Proof-of-Concept Study of Using Supervised Machine Learning Algorithms to Predict Self-Care and Glycemic Control in Type 1 Diabetes Patients on Insulin Pump Therapy

ObjectiveUsing supervised machine learning algorithms (SMLAs), we built models to predict the probability of type 1 diabetes mellitus patients on insulin pump therapy for meeting insulin pump self-management behavioral (IPSMB) criteria and achieving good glycemic response within 6 months.MethodsThis was a single-center retrospective chart review of 100 adult type 1 diabetes mellitus patients on insulin pump therapy (≥6 months). Three SMLAs were deployed: multivariable logistic regression (LR), random forest (RF), and K-nearest neighbor (k-NN); validated using repeated three-fold cross-validation. Performance metrics included area under the curve-Receiver of characteristics for discrimination and Brier scores for calibration.ResultsVariables predictive of adherence with IPSMB criteria were baseline hemoglobin A1c, continuous glucose monitoring, and sex. The models had comparable discriminatory power (LR = 0.74; RF = 0.74; k-NN = 0.72), with the RF model showing better calibration (Brier = 0.151). Predictors of the good glycemic response included baseline hemoglobin A1c, entering carbohydrates, and following the recommended bolus dose, with models comparable in discriminatory power (LR = 0.81, RF = 0.80, k-NN = 0.78) but the RF model being better calibrated (Brier = 0.099).ConclusionThese proof-of-concept analyses demonstrate the feasibility of using SMLAs to develop clinically relevant predictive models of adherence with IPSMB criteria and glycemic control within 6 months. Subject to further study, nonlinear prediction models may perform better.     

A Deep Feature Learning Model for Pneumonia Detection Applying a Combination of mRMR Feature Selection and Machine Learning Models

Pneumonia is one of the diseases that people may encounter in any period of their lives. Approximately 18% of infectious diseases are caused by pneumonia. This disease may result in death in the following stages. In order to diagnose pneumonia as a medical condition, lung X-ray images are routinely examined by the field experts in the clinical practice. In this study, lung X-ray images that are available for the diagnosis of pneumonia were used. The convolutional neural network was employed as feature extractor, and some of existing convolutional neural network models that are AlexNet, VGG-16 and VGG-19 were utilized so as to realize this specific task. Then, the number of deep features was reduced from 1000 to 100 by using the minimum redundancy maximum relevance algorithm for each deep model. Accordingly, we achieved 100 deep features from each deep model, and we combined these features so as to provide an efficient feature set consisting of totally 300 deep features. In this step of the experiment, this feature set was given as an input to the decision tree, k-nearest neighbors, linear discriminant analysis, linear regression, and support vector machine learning models. Finally, all models ensured promising results, especially linear discriminant analysis yielded the most efficient results with an accuracy of 99.41%. Consequently, the results point out that the deep features provided robust and consistent features for pneumonia detection, and minimum redundancy maximum relevance method was found a beneficial tool to reduce the dimension of the feature set.     

Extreme Learning Machine-Based Classification of ADHD Using Brain Structural MRI Data

Background

Effective and accurate diagnosis of attention-deficit/hyperactivity disorder (ADHD) is currently of significant interest. ADHD has been associated with multiple cortical features from structural MRI data. However, most existing learning algorithms for ADHD identification contain obvious defects, such as time-consuming training, parameters selection, etc. The aims of this study were as follows: (1) Propose an ADHD classification model using the extreme learning machine (ELM) algorithm for automatic, efficient and objective clinical ADHD diagnosis. (2) Assess the computational efficiency and the effect of sample size on both ELM and support vector machine (SVM) methods and analyze which brain segments are involved in ADHD.

Methods

High-resolution three-dimensional MR images were acquired from 55 ADHD subjects and 55 healthy controls. Multiple brain measures (cortical thickness, etc.) were calculated using a fully automated procedure in the FreeSurfer software package. In total, 340 cortical features were automatically extracted from 68 brain segments with 5 basic cortical features. F-score and SFS methods were adopted to select the optimal features for ADHD classification. Both ELM and SVM were evaluated for classification accuracy using leave-one-out cross-validation.

Results

We achieved ADHD prediction accuracies of 90.18% for ELM using eleven combined features, 84.73% for SVM-Linear and 86.55% for SVM-RBF. Our results show that ELM has better computational efficiency and is more robust as sample size changes than is SVM for ADHD classification. The most pronounced differences between ADHD and healthy subjects were observed in the frontal lobe, temporal lobe, occipital lobe and insular.

Conclusion

Our ELM-based algorithm for ADHD diagnosis performs considerably better than the traditional SVM algorithm. This result suggests that ELM may be used for the clinical diagnosis of ADHD and the investigation of different brain diseases.     

HMM-based Supervised Machine Learning Framework for the Detection of fECG R-R Peak Locations

ObjectiveFetal Electro Cardiogram (fECG) provides critical information on the wellbeing of a foetus heart in its developing stages in the mother's womb. The objective of this work is to extract fECG which is buried in a composite signal consisting of itself, maternal ECG (mECG) and noises contributed from various unavoidable sources. In the past, the challenge of extracting fECG from the composite signal was dealt with by Stochastic Weiner filter, model-based Kalman filter and other adaptive filtering techniques. Blind Source Separation (BSS) based Independent Component Analysis (ICA) has shown an edge over the adaptive filtering techniques as the former does not require a reference signal. Recently, data-driven machine learning techniques e.g., adaptive neural networks, adaptive neuro-fuzzy inference system, support vector machine (SVM) are also applied.MethodThis work pursues hidden Markov model (HMM)-based supervised machine learning frame-work for the determination of the location of fECG QRS complex from the composite abdominal signal. HMM is used to model the underlying hidden states of the observable time series of the extracted and separated fECG data with its QRS peak location as one of the hidden states. The state transition probabilities are estimated in the training phase using the annotated data sets. Afterwards, using the estimated HMM networks, fQRS locations are detected in the testing phase. To evaluate the proposed technique, the accuracy of the correct detection of QRS complex with respect to the correct annotation of QRS complex location is considered and quantified by the sensitivity, probability of false alarm, and accuracy.ResultsThe best results that have been achieved using the proposed method are: accuracy – 97.1%, correct detection rate (translated to sensitivity) – 100%, and false alarm rate – 2.89%.ConclusionTwo primary challenges in these methods are finding the right reference threshold for the normalization of the extracted fECG signal during the initial trials and limitation of discrete frame work of HMM signal (converted from continuous time) which only offers a countable number of levels in observations. By feeding the posterior probabilities, obtained from SVM, into HMM, as emission probabilities, can further improve the accuracy of fQRS location detection.     

A Hybrid Deep Learning Model for Predicting Molecular Subtypes of Human Breast Cancer Using Multimodal Data

BackgroundThe prediction of breast cancer subtypes plays a key role in the diagnosis and prognosis of breast cancer. In recent years, deep learning (DL) has shown good performance in the intelligent prediction of breast cancer subtypes. However, most of the traditional DL models use single modality data, which can just extract a few features, so it cannot establish a stable relationship between patient characteristics and breast cancer subtypes.DatasetWe used the TCGA-BRCA dataset as a sample set for molecular subtype prediction of breast cancer. It is a public dataset that can be obtained through the following link: https://portal.gdc.cancer.gov/projects/TCGA-BRCAMethodsIn this paper, a Hybrid DL model based on the multimodal data is proposed. We combine the patient's gene modality data with image modality data to construct a multimodal fusion framework. According to the different forms and states, we set up feature extraction networks respectively, and then we fuse the output of the two feature networks based on the idea of weighted linear aggregation. Finally, the fused features are used to predict breast cancer subtypes. In particular, we use the principal component analysis to reduce the dimensionality of high-dimensional data of gene modality and filter the data of image modality. Besides, we also improve the traditional feature extraction network to make it show better performance.ResultsThe results show that compared with the traditional DL model, the Hybrid DL model proposed in this paper is more accurate and efficient in predicting breast cancer subtypes. Our model achieved a prediction accuracy of 88.07% in 10 times of 10-fold cross-validation. We did a separate AUC test for each subtype, and the average AUC value obtained was 0.9427. In terms of subtype prediction accuracy, our model is about 7.45% higher than the previous average.     

Cerebral Blood Volume Analysis in Glioblastomas Using Dynamic Susceptibility Contrast-Enhanced Perfusion MRI: A Comparison of Manual and Semiautomatic Segmentation Methods

Purpose

To compare the reproducibilities of manual and semiautomatic segmentation method for the measurement of normalized cerebral blood volume (nCBV) using dynamic susceptibility contrast-enhanced (DSC) perfusion MR imaging in glioblastomas.

Materials and Methods

Twenty-two patients (11 male, 11 female; 27 tumors) with histologically confirmed glioblastoma (WHO grade IV) were examined with conventional MR imaging and DSC imaging at 3T before surgery or biopsy. Then nCBV (means and standard deviations) in each mass was measured using two DSC MR perfusion analysis methods including manual and semiautomatic segmentation method, in which contrast-enhanced (CE)-T1WI and T2WI were used as structural imaging. Intraobserver and interobserver reproducibility were assessed according to each perfusion analysis method or each structural imaging. Interclass correlation coefficient (ICC), Bland-Altman plot, and coefficient of variation (CV) were used to evaluate reproducibility.

Results

Intraobserver reproducibilities on CE-T1WI and T2WI were ICC of 0.74–0.89 and CV of 20.39–36.83% in manual segmentation method, and ICC of 0.95–0.99 and CV of 8.53–16.19% in semiautomatic segmentation method, repectively. Interobserver reproducibilites on CE-T1WI and T2WI were ICC of 0.86–0.94 and CV of 19.67–35.15% in manual segmentation method, and ICC of 0.74–1.0 and CV of 5.48–49.38% in semiautomatic segmentation method, respectively. Bland-Altman plots showed a good correlation with ICC or CV in each method. The semiautomatic segmentation method showed higher intraobserver and interobserver reproducibilities at CE-T1WI-based study than other methods.

Conclusion

The best reproducibility was found using the semiautomatic segmentation method based on CE-T1WI for structural imaging in the measurement of the nCBV of glioblastomas.     

Automatic Detection of Severely and Mildly Infected COVID-19 Patients with Supervised Machine Learning Models

ObjectivesWhen the prognosis of COVID-19 disease can be detected early, the intense-pressure and loss of workforce in health-services can be partially reduced. The primary-purpose of this article is to determine the feature-dataset consisting of the routine-blood-values (RBV) and demographic-data that affect the prognosis of COVID-19. Second, by applying the feature-dataset to the supervised machine-learning (ML) models, it is to identify severely and mildly infected COVID-19 patients at the time of admission.Material and methodsThe sample of this study consists of severely (n = 192) and mildly (n = 4010) infected-patients hospitalized with the diagnosis of COVID-19 between March-September, 2021. The RBV-data measured at the time of admission and age-gender characteristics of these patients were analyzed retrospectively. For the selection of the features, the minimum-redundancy-maximum-relevance (MRMR) method, principal-components-analysis and forward-multiple-logistics-regression analyzes were used. The features set were statistically compared between mild and severe infected-patients. Then, the performances of various supervised-ML-models were compared in identifying severely and mildly infected-patients using the feature set.ResultsIn this study, 28 RBV-parameters and age-variable were found as the feature-dataset. The effect of features on the prognosis of the disease has been clinically proven. The ML-models with the highest overall-accuracy in identifying patient-groups were found respectively, as follows: local-weighted-learning (LWL)-97.86%, K-star (K*)-96.31%, Naive-Bayes (NB)-95.36% and k-nearest-neighbor (KNN)-94.05%. Also, the most successful models with the highest area-under-the-receiver-operating-characteristic-curve (AUC) values in identifying patient groups were found respectively, as follows: LWL-0.95%, K*-0.91%, NB-0.85% and KNN-0.75%.ConclusionThe findings in this article have significant a motivation for the healthcare professionals to detect at admission severely and mildly infected COVID-19 patients.     

Alterations of Gray and White Matter Networks in Patients with Obsessive-Compulsive Disorder: A Multimodal Fusion Analysis of Structural MRI and DTI Using mCCA+jICA

Many of previous neuroimaging studies on neuronal structures in patients with obsessive-compulsive disorder (OCD) used univariate statistical tests on unimodal imaging measurements. Although the univariate methods revealed important aberrance of local morphometry in OCD patients, the covariance structure of the anatomical alterations remains unclear. Motivated by recent developments of multivariate techniques in the neuroimaging field, we applied a fusion method called “mCCA+jICA” on multimodal structural data of T1-weighted magnetic resonance imaging (MRI) and diffusion tensor imaging (DTI) of 30 unmedicated patients with OCD and 34 healthy controls. Amongst six highly correlated multimodal networks (p < 0.0001), we found significant alterations of the interrelated gray and white matter networks over occipital and parietal cortices, frontal interhemispheric connections and cerebella (False Discovery Rate q ≤ 0.05). In addition, we found white matter networks around basal ganglia that correlated with a subdimension of OC symptoms, namely ‘harm/checking’ (q ≤ 0.05). The present study not only agrees with the previous unimodal findings of OCD, but also quantifies the association of the altered networks across imaging modalities.     

A Comparison of Supervised Classification Methods for the Prediction of Substrate Type Using Multibeam Acoustic and Legacy Grain-Size Data

Detailed seabed substrate maps are increasingly in demand for effective planning and management of marine ecosystems and resources. It has become common to use remotely sensed multibeam echosounder data in the form of bathymetry and acoustic backscatter in conjunction with ground-truth sampling data to inform the mapping of seabed substrates. Whilst, until recently, such data sets have typically been classified by expert interpretation, it is now obvious that more objective, faster and repeatable methods of seabed classification are required. This study compares the performances of a range of supervised classification techniques for predicting substrate type from multibeam echosounder data. The study area is located in the North Sea, off the north-east coast of England. A total of 258 ground-truth samples were classified into four substrate classes. Multibeam bathymetry and backscatter data, and a range of secondary features derived from these datasets were used in this study. Six supervised classification techniques were tested: Classification Trees, Support Vector Machines, k-Nearest Neighbour, Neural Networks, Random Forest and Naive Bayes. Each classifier was trained multiple times using different input features, including i) the two primary features of bathymetry and backscatter, ii) a subset of the features chosen by a feature selection process and iii) all of the input features. The predictive performances of the models were validated using a separate test set of ground-truth samples. The statistical significance of model performances relative to a simple baseline model (Nearest Neighbour predictions on bathymetry and backscatter) were tested to assess the benefits of using more sophisticated approaches. The best performing models were tree based methods and Naive Bayes which achieved accuracies of around 0.8 and kappa coefficients of up to 0.5 on the test set. The models that used all input features didn''t generally perform well, highlighting the need for some means of feature selection.     

Comparison of Algorithms for Prediction of Protein Structural Features from Evolutionary Data

Proteins have many functions and predicting these is still one of the major challenges in theoretical biophysics and bioinformatics. Foremost amongst these functions is the need to fold correctly thereby allowing the other genetically dictated tasks that the protein has to carry out to proceed efficiently. In this work, some earlier algorithms for predicting protein domain folds are revisited and they are compared with more recently developed methods. In dealing with intractable problems such as fold prediction, when different algorithms show convergence onto the same result there is every reason to take all algorithms into account such that a consensus result can be arrived at. In this work it is shown that the application of different algorithms in protein structure prediction leads to results that do not converge as such but rather they collude in a striking and useful way that has never been considered before.     

Fast and Accurate Bacterial Species Identification in Urine Specimens Using LC-MS/MS Mass Spectrometry and Machine Learning*

1. Download : Download high-res image (93KB)
2. Download : Download full-size image

Highlights

• •Fast and culture-free method for the identification of the 15 bacterial species causing UTIs.
• •Combination of DIA analysis and machine learning algorithms to define a peptide signature.
• •High accuracy, good linearity and reproducibility, sensitivity below standard threshold.
• •Transferability to other laboratories and other mass spectrometers.

     

Fully Bayesian Inference for Structural MRI: Application to Segmentation and Statistical Analysis of T2-Hypointensities

Aiming at iron-related T2-hypointensity, which is related to normal aging and neurodegenerative processes, we here present two practicable approaches, based on Bayesian inference, for preprocessing and statistical analysis of a complex set of structural MRI data. In particular, Markov Chain Monte Carlo methods were used to simulate posterior distributions. First, we rendered a segmentation algorithm that uses outlier detection based on model checking techniques within a Bayesian mixture model. Second, we rendered an analytical tool comprising a Bayesian regression model with smoothness priors (in the form of Gaussian Markov random fields) mitigating the necessity to smooth data prior to statistical analysis. For validation, we used simulated data and MRI data of 27 healthy controls (age: ; range, ). We first observed robust segmentation of both simulated T2-hypointensities and gray-matter regions known to be T2-hypointense. Second, simulated data and images of segmented T2-hypointensity were analyzed. We found not only robust identification of simulated effects but also a biologically plausible age-related increase of T2-hypointensity primarily within the dentate nucleus but also within the globus pallidus, substantia nigra, and red nucleus. Our results indicate that fully Bayesian inference can successfully be applied for preprocessing and statistical analysis of structural MRI data.