Enhanced manifold regularization for semi-supervised classification.

Manifold regularization (MR) has become one of the most widely used approaches in the semi-supervised learning field. It has shown superiority by exploiting the local manifold structure of both labeled and unlabeled data. The manifold structure is modeled by constructing a Laplacian graph and then incorporated in learning through a smoothness regularization term. Hence the labels of labeled and unlabeled data vary smoothly along the geodesics on the manifold. However, MR has ignored the discriminative ability of the labeled and unlabeled data. To address the problem, we propose an enhanced MR framework for semi-supervised classification in which the local discriminative information of the labeled and unlabeled data is explicitly exploited. To make full use of labeled data, we firstly employ a semi-supervised clustering method to discover the underlying data space structure of the whole dataset. Then we construct a local discrimination graph to model the discriminative information of labeled and unlabeled data according to the discovered intrinsic structure. Therefore, the data points that may be from different clusters, though similar on the manifold, are enforced far away from each other. Finally, the discrimination graph is incorporated into the MR framework. In particular, we utilize semi-supervised fuzzy c-means and Laplacian regularized Kernel minimum squared error for semi-supervised clustering and classification, respectively. Experimental results on several benchmark datasets and face recognition demonstrate the effectiveness of our proposed method.

Scale-Dependent Signal Identification in Low-Dimensional Subspace: Motor Imagery Task Classification.

Motor imagery electroencephalography (EEG) has been successfully used in locomotor rehabilitation programs. While the noise-assisted multivariate empirical mode decomposition (NA-MEMD) algorithm has been utilized to extract task-specific frequency bands from all channels in the same scale as the intrinsic mode functions (IMFs), identifying and extracting the specific IMFs that contain significant information remain difficult. In this paper, a novel method has been developed to identify the information-bearing components in a low-dimensional subspace without prior knowledge. Our method trains a Gaussian mixture model (GMM) of the composite data, which is comprised of the IMFs from both the original signal and noise, by employing kernel spectral regression to reduce the dimension of the composite data. The informative IMFs are then discriminated using a GMM clustering algorithm, the common spatial pattern (CSP) approach is exploited to extract the task-related features from the reconstructed signals, and a support vector machine (SVM) is applied to the extracted features to recognize the classes of EEG signals during different motor imagery tasks. The effectiveness of the proposed method has been verified by both computer simulations and motor imagery EEG datasets.

Manifold regularized semi-supervised Gaussian mixture model.

In the last decades, Gaussian Mixture Models (GMMs) have attracted considerable interest in data mining and pattern recognition. A GMM-based clustering algorithm models a dataset with a mixture of multiple Gaussian components and estimates the model parameters using the Expectation-Maximization (EM) algorithm. Recently, a new Locally Consistent GMM (LCGMM) has been proposed to improve the clustering performance by exploiting the local manifold structure of the data using a p nearest neighbor graph. In addition to the underlying manifold structure, many other forms of prior knowledge may guide the clustering process and improve the performance. In this paper, we introduce a Semi-Supervised LCGMM (Semi-LCGMM), where the prior knowledge is provided in the form of class labels of partial data. In particular, the new Semi-LCGMM incorporates the prior knowledge into the maximum likelihood function of the original LCGMM, and the model parameters are estimated using the EM algorithm. It is worth noting that, in our algorithm, each class may be modeled by multiple Gaussian components while in the unsupervised setting each class is modeled by a single Gaussian component. Our algorithm has shown promising results in many different applications, including clustering breast cancer data, heart disease data, handwritten digit images, human face images, and image segmentation.

Action recognition through discovering distinctive action parts.

Recent methods based on midlevel visual concepts have shown promising capabilities in the human action recognition field. Automatically discovering semantic entities such as action parts remains challenging. In this paper, we present a method of automatically discovering distinctive midlevel action parts from video for recognition of human actions. We address this problem by learning and selecting a collection of discriminative and representative action part detectors directly from video data. We initially train a large collection of candidate exemplar-linear discriminant analysis detectors from clusters obtained by clustering spatiotemporal patches in whitened space. To select the most effective detectors from the vast array of candidates, we propose novel coverage-entropy curves (CE curves) to evaluate a detector's capability of distinguishing actions. The CE curves characterize the correlation between the representative and discriminative power of detectors. In the experiments, we apply the mined part detectors as a visual vocabulary to the task of action recognition on four datasets: KTH, Olympic Sports, UCF50, and HMDB51. The experimental results demonstrate the effectiveness of the proposed method and show the state-of-the-art recognition performance.

Self-training-based face recognition using semi-supervised linear discriminant analysis and affinity propagation.

Face recognition is one of the most important applications of machine learning and computer vision. The traditional supervised learning methods require a large amount of labeled face images to achieve good performance. In practice, however, labeled images are usually scarce while unlabeled ones may be abundant. In this paper, we introduce a semi-supervised face recognition method, in which semi-supervised linear discriminant analysis (SDA) and affinity propagation (AP) are integrated into a self-training framework. In particular, SDA is employed to compute the face subspace using both labeled and unlabeled images, and AP is used to identify the exemplars of different face classes in the subspace. The unlabeled data can then be classified according to the exemplars and the newly labeled data with the highest confidence are added to the labeled data, and the whole procedure iterates until convergence. A series of experiments on four face datasets are carried out to evaluate the performance of our algorithm. Experimental results illustrate that our algorithm outperforms the other unsupervised, semi-supervised, and supervised methods.

An Alzheimer's Disease classification network based on MRI utilizing diffusion maps for multi-scale feature fusion in graph convolution.

Graph convolutional networks (GCN) have been widely utilized in Alzheimer's disease (AD) classification research due to its ability to automatically learn robust and powerful feature representations. Inter-patient relationships are effectively captured by constructing patients magnetic resonance imaging (MRI) data as graph data, where nodes represent individuals and edges denote the relationships between them. However, the performance of GCNs might be constrained by the construction of the graph adjacency matrix, thereby leading to learned features potentially overlooking intrinsic correlations among patients, which ultimately causes inaccurate disease classifications. To address this issue, we propose an Alzheimer's disease Classification network based on MRI utilizing diffusion maps for multi-scale feature fusion in graph convolution. This method aims to tackle the problem of features neglecting intrinsic relationships among patients while integrating features from diffusion mapping with different neighbor counts to better represent patients and achieve an accurate AD classification. Initially, the diffusion maps method conducts diffusion information in the feature space, thus breaking free from the constraints of diffusion based on the adjacency matrix. Subsequently, the diffusion features with different neighbor counts are merged, and a self-attention mechanism is employed to adaptively adjust the weights of diffusion features at different scales, thereby comprehensively and accurately capturing patient characteristics. Finally, metric learning techniques enhance the similarity of node features within the same category in the graph structure and bring node features of different categories more distant from each other. This study aims to enhance the classification accuracy of AD, by providing an effective tool for early diagnosis and intervention. It offers valuable information for clinical decisions and personalized treatment. Experimentation on the publicly accessible Alzheimer's disease neuroimaging initiative (ADNI) dataset validated our method's competitive performance across various AD-related classification tasks. Compared to existing methodologies, our approach captures patient characteristics more effectively and demonstrates superior generalization capabilities.

Dual uncertainty-guided multi-model pseudo-label learning for semi-supervised medical image segmentation.

Semi-supervised medical image segmentation is currently a highly researched area. Pseudo-label learning is a traditional semi-supervised learning method aimed at acquiring additional knowledge by generating pseudo-labels for unlabeled data. However, this method relies on the quality of pseudo-labels and can lead to an unstable training process due to differences between samples. Additionally, directly generating pseudo-labels from the model itself accelerates noise accumulation, resulting in low-confidence pseudo-labels. To address these issues, we proposed a dual uncertainty-guided multi-model pseudo-label learning framework (DUMM) for semi-supervised medical image segmentation. The framework consisted of two main parts: The first part is a sample selection module based on sample-level uncertainty (SUS), intended to achieve a more stable and smooth training process. The second part is a multi-model pseudo-label generation module based on pixel-level uncertainty (PUM), intended to obtain high-quality pseudo-labels. We conducted a series of experiments on two public medical datasets, ACDC2017 and ISIC2018. Compared to the baseline, we improved the Dice scores by 6.5% and 4.0% over the two datasets, respectively. Furthermore, our results showed a clear advantage over the comparative methods. This validates the feasibility and applicability of our approach.

An image registration-based self-supervised Su-Net for carotid plaque ultrasound image segmentation.

BACKGROUND AND OBJECTIVES: Total Plaque Area (TPA) measurement is critical for early diagnosis and intervention of carotid atherosclerosis in individuals with high risk for stroke. The delineation of the carotid plaques is necessary for TPA measurement, and deep learning methods can automatically segment the plaque and measure TPA from carotid ultrasound images. A large number of labeled images is essential for training a good deep learning model, but it is very difficult to collect such large labeled datasets for carotid image segmentation in clinical practice. Self-supervised learning can provide a possible solution to improve the deep-learning models on small labeled training datasets by designing a pretext task to pre-train the models without using the segmentation masks. However, the existing self-supervised learning methods do not consider the feature presentations of object contours. METHODS: In this paper, we propose an image registration-based self-supervised learning method and a stacked U-Net (SSL-SU-Net) for carotid plaque ultrasound image segmentation, which can better exploit the semantic features of carotid plaque contours in self-supervised task training. RESULTS: Our network was trained on different numbers of labeled images (n = 10, 33, 50 and 100 subjects) and tested on 44 subjects from the SPARC dataset (n = 144, London, Canada). The network trained on the entire SPARC dataset was then directly applied to an independent dataset collected in Zhongnan hospital (n = 497, Wuhan, China). For the 44 subjects tested on the SPARC dataset, our method yielded a DSC of 80.25-89.18% and the produced TPA measurements, which were strongly correlated with manual segmentation (r = 0.965-0.995, ρ< 0.0001). For the Zhongnan dataset, the DSC was 90.3% and algorithm TPAs were strongly correlated with manual TPAs (r = 0.985, ρ< 0.0001). CONCLUSIONS: The results demonstrate that our proposed method yielded excellent performance and good generalization ability when trained on a small labeled dataset, facilitating the use of deep learning in carotid ultrasound image analysis and clinical practice. The code of our algorithm is available https://github.com/a610lab/Registration-SSL.

A self-supervised fusion network for carotid plaque ultrasound image classification.

Carotid plaque classification from ultrasound images is crucial for predicting ischemic stroke risk. While deep learning has shown effectiveness, it heavily relies on substantial labeled datasets. Achieving high performance with limited labeled images is essential for clinical use. Self-supervised learning (SSL) offers a potential solution; however, the existing works mainly focus on constructing the SSL tasks, neglecting the use of multiple tasks for pretraining. To overcome these limitations, this study proposed a self-supervised fusion network (Fusion-SSL) for carotid plaque ultrasound image classification with limited labeled data. Fusion-SSL consists of two SSL tasks: classifying image block order (Ordering) and predicting image rotation angle (Rotating). A dual-branch residual neural network was developed to fuse feature presentations learned by the two tasks, which can extract richer visual boundary shape and contour information than a single task. In this experiment, 1270 carotid plaque ultrasound images were collected from 844 patients at Zhongnan Hospital (Wuhan, China). The results showed that Fusion-SSL outperforms single SSL methods across different percentages of labeled training data, ranging from 10 to 100%. Moreover, with only 40% labeled training data, Fusion-SSL achieved comparable results to a single SSL method (predicting image rotation angle) with 100% labeled data. These results indicate that Fusion-SSL could be beneficial for the classification of carotid plaques and the early warning of a stroke in clinical practice.

JoCoRank: Joint correlation learning with ranking similarity regularization for imbalanced fetal brain age regression.

Estimating fetal brain age based on sulci by magnetic resonance imaging (MRI) is clinically crucial in determining the normal development of fetal brains. Deep learning provides a possible way for fetal brain age estimation using MRI. Previous studies have mainly emphasized optimizing individual-wise correlation criteria, such as mean square error. However, they ignored the very important global and peer-wise criterion, which are essential for learning the structured relationships among regression samples. Moreover, the imbalanced label distribution introduces an adverse bias, which impairs the reliability and interpretation of correlation estimation and the model's fairness and generalizability. In this work, we propose a novel joint correlation learning with ranking similarity regularization (JoCoRank) algorithm for deep imbalanced regression of fetal brain age. Joint correlation learning concurrently captures individual, global, and peer-level valuable relationship information, and the customized optimization scheme for each criterion exhibits strong robustness against outliers and imbalanced regression. Ranking similarity regularization is designed to calibrate the biased feature representations by aligning the sorted list of neighbors in the label space with those in the feature space. A total of 1327 MRI images from 157 healthy fetuses between 22 and 34 weeks were collected at Wuhan Children's Hospital and utilized to evaluate the performance of JoCoRank in fetal brain age estimation. JoCoRank achieved promising results with an average mean absolute error of 0.693±0.064 weeks and R2 coefficient of 0.930±0.019. Our fetal brain age estimation algorithm would be useful for identifying abnormalities in fetal brain development and undertaking early intervention in clinical practice.

Mutual-DTI: A mutual interaction feature-based neural network for drug-target protein interaction prediction.

The prediction of drug-target protein interaction (DTI) is a crucial task in the development of new drugs in modern medicine. Accurately identifying DTI through computer simulations can significantly reduce development time and costs. In recent years, many sequence-based DTI prediction methods have been proposed, and introducing attention mechanisms has improved their forecasting performance. However, these methods have some shortcomings. For example, inappropriate dataset partitioning during data preprocessing can lead to overly optimistic prediction results. Additionally, only single non-covalent intermolecular interactions are considered in the DTI simulation, ignoring the complex interactions between their internal atoms and amino acids. In this paper, we propose a network model called Mutual-DTI that predicts DTI based on the interaction properties of sequences and a Transformer model. We use multi-head attention to extract the long-distance interdependent features of the sequence and introduce a module to extract the sequence's mutual interaction features in mining complex reaction processes of atoms and amino acids. We evaluate the experiments on two benchmark datasets, and the results show that Mutual-DTI outperforms the latest baseline significantly. In addition, we conduct ablation experiments on a label-inversion dataset that is split more rigorously. The results show that there is a significant improvement in the evaluation metrics after introducing the extracted sequence interaction feature module. This suggests that Mutual-DTI may contribute to modern medical drug development research. The experimental results show the effectiveness of our approach. The code for Mutual-DTI can be downloaded from https://github.com/a610lab/Mutual-DTI.

Carotid Vessel-Wall-Volume Ultrasound Measurement via a UNet++ Ensemble Algorithm Trained on Small Data Sets.

Vessel wall volume (VWV) is a 3-D ultrasound measurement for the assessment of therapy in patients with carotid atherosclerosis. Deep learning can be used to segment the media-adventitia boundary (MAB) and lumen-intima boundary (LIB) and to quantify VWV automatically; however, it typically requires large training data sets with expert manual segmentation, which are difficult to obtain. In this study, a UNet++ ensemble approach was developed for automated VWV measurement, trained on five small data sets (n = 30 participants) and tested on 100 participants with clinically diagnosed coronary artery disease enrolled in a multicenter CAIN trial. The Dice similarity coefficient (DSC), average symmetric surface distance (ASSD), Pearson correlation coefficient (r), Bland-Altman plots and coefficient of variation (CoV) were used to evaluate algorithm segmentation accuracy, agreement and reproducibility. The UNet++ ensemble yielded DSCs of 91.07%-91.56% and 87.53%-89.44% and ASSDs of 0.10-0.11 mm and 0.33-0.39 mm for the MAB and LIB, respectively; the algorithm VWV measurements were correlated (r = 0.763-0.795, p < 0.001) with manual segmentations, and the CoV for VWV was 8.89%. In addition, the UNet++ ensemble trained on 30 participants achieved a performance similar to that of U-Net and Voxel-FCN trained on 150 participants. These results suggest that our approach could provide accurate and reproducible carotid VWV measurements using relatively small training data sets, supporting deep learning applications for monitoring atherosclerosis progression in research and clinical trials.

PDFF-CNN: An attention-guided dynamic multi-orientation feature fusion method for gestational age prediction on imbalanced fetal brain MRI dataset.

BACKGROUND: Fetal brain magnetic resonance imaging (MRI)-based gestational age prediction has been widely used to characterize normal fetal brain development and diagnose congenital brain malformations. PURPOSE: The uncertainty of fetal position and external interference leads to variable localization and direction of the fetal brain. In addition, pregnant women typically concentrate on receiving MRI scans during the fetal anomaly scanning week, leading to an imbalanced distribution of fetal brain MRI data. The above-mentioned problems pose great challenges for deep learning-based fetal brain MRI gestational age prediction. METHODS: In this study, a pyramid squeeze attention (PSA)-guided dynamic feature fusion CNN (PDFF-CNN) is proposed to robustly predict gestational ages from fetal brain MRI images on an imbalanced dataset. PDFF-CNN contains four components: transformation module, feature extraction module, dynamic feature fusion module, and balanced mean square error (MSE) loss. The transformation and feature extraction modules are employed by using the PSA to learn multiscale and multi-orientation feature representations in a parallel weight-sharing Siamese network. The dynamic feature fusion module automatically learns the weights of feature vectors generated in the feature extraction module to dynamically fuse multiscale and multi-orientation brain sulci and gyri features. Considering the fact of the imbalanced dataset, the balanced MSE loss is used to mitigate the negative impact of imbalanced data distribution on gestational age prediction performance. RESULTS: Evaluated on an imbalanced fetal brain MRI dataset of 1327 routine clinical T2-weighted MRI images from 157 subjects, PDFF-CNN achieved promising gestational age prediction performance with an overall mean absolute error of 0.848 weeks and an R2 of 0.904. Furthermore, the attention activation maps of PDFF-CNN were derived, which revealed regional features that contributed to gestational age prediction at each gestational stage. CONCLUSIONS: These results suggest that the proposed PDFF-CNN might have broad clinical applicability in guiding treatment interventions and delivery planning, which has the potential to be helpful with prenatal diagnosis.

Ultrasound carotid plaque segmentation via image reconstruction-based self-supervised learning with limited training labels.

Carotid total plaque area (TPA) is an important contributing measurement to the evaluation of stroke risk. Deep learning provides an efficient method for ultrasound carotid plaque segmentation and TPA quantification. However, high performance of deep learning requires datasets with many labeled images for training, which is very labor-intensive. Thus, we propose an image reconstruction-based self-supervised learning algorithm (IR-SSL) for carotid plaque segmentation when few labeled images are available. IR-SSL consists of pre-trained and downstream segmentation tasks. The pre-trained task learns region-wise representations with local consistency by reconstructing plaque images from randomly partitioned and disordered images. The pre-trained model is then transferred to the segmentation network as the initial parameters in the downstream task. IR-SSL was implemented with two networks, UNet++ and U-Net, and evaluated on two independent datasets of 510 carotid ultrasound images from 144 subjects at SPARC (London, Canada) and 638 images from 479 subjects at Zhongnan hospital (Wuhan, China). Compared to the baseline networks, IR-SSL improved the segmentation performance when trained on few labeled images (n = 10, 30, 50 and 100 subjects). For 44 SPARC subjects, IR-SSL yielded Dice-similarity-coefficients (DSC) of 80.14-88.84%, and algorithm TPAs were strongly correlated (r=0.962-0.993, p < 0.001) with manual results. The models trained on the SPARC images but applied to the Zhongnan dataset without retraining achieved DSCs of 80.61-88.18% and strong correlation with manual segmentation (r=0.852-0.978, p < 0.001). These results suggest that IR-SSL could improve deep learning when trained on small labeled datasets, making it useful for monitoring carotid plaque progression/regression in clinical use and trials.

An improved self-supervised learning for EEG classification.

Motor Imagery EEG (MI-EEG) classification plays an important role in different Brain-Computer Interface (BCI) systems. Recently, deep learning has been widely used in the MI-EEG classification tasks, however this technology requires a large number of labeled training samples which are difficult to obtain, and insufficient labeled training samples will result in a degradation of the classification performance. To address the degradation problem, we investigate a Self-Supervised Learning (SSL) based MI-EEG classification method to reduce the dependence on a large number of labeled training samples. The proposed method includes a pretext task and a downstream classification one. In the pretext task, each MI-EEG is rearranged according to the temporal characteristic. A network is pre-trained using the original and rearranged MI-EEGs. In the downstream task, a MI-EEG classification network is firstly initialized by the network learned in the pretext task and then trained using a small number of the labeled training samples. A series of experiments are conducted on Data sets 1 and 2b of BCI competition IV and IVa of BCI competition III. In the case of one third of the labeled training samples, the proposed method can obtain an obvious improvement compared to the baseline network without using SSL. In the experiments under different percentages of the labeled training samples, the results show that the designed SSL strategy is effective and beneficial to improving the classification performance.

Time-of-flight completion in ultrasound computed tomography based on the singular value threshold algorithm.

Ultrasound computed tomography (USCT) has been developed for breast tumor screening. The sound-speed modal of USCT can provide quantitative sound-speed values to help tumor diagnosis. Time-of-flight (TOF) is the critical input in sound-speed reconstruction. However, we found that the missing data problem in the detected TOF causes artifacts on the reconstructed sound-speed images, which may affect the tumor identification. In this study, to address the missing TOF data problem, we first adopted the singular value threshold (SVT) algorithm to complete the TOF matrix. The threshold value in SVT is difficult to determine, so we proposed a selection strategy, that is, to enumerate the threshold values as the multiples of the maximum singular value of the incomplete matrix and then evaluate the image quality to select the proper threshold value. In the numerical breast phantom experiment, the artifacts are eliminated, and the accuracy is higher than the accuracy of the compared methods. In the in vivo experiment, we reconstructed the sound-speed image of the breast of a volunteer with invasive breast cancer, and the SVT algorithm improved the image sharpness. The completion of DTOF based on SVT gives better accuracy than the compared methods, but too large a threshold value decreases the accuracy. In the future, the selection method of the threshold value needs further research, and more USCT cases should be included in the experiments.

A safe semi-supervised graph convolution network.

In the semi-supervised learning field, Graph Convolution Network (GCN), as a variant model of GNN, has achieved promising results for non-Euclidean data by introducing convolution into GNN. However, GCN and its variant models fail to safely use the information of risk unlabeled data, which will degrade the performance of semi-supervised learning. Therefore, we propose a Safe GCN framework (Safe-GCN) to improve the learning performance. In the Safe-GCN, we design an iterative process to label the unlabeled data. In each iteration, a GCN and its supervised version (S-GCN) are learned to find the unlabeled data with high confidence. The high-confidence unlabeled data and their pseudo labels are then added to the label set. Finally, both added unlabeled data and labeled ones are used to train a S-GCN which can achieve the safe exploration of the risk unlabeled data and enable safe use of large numbers of unlabeled data. The performance of Safe-GCN is evaluated on three well-known citation network datasets and the obtained results demonstrate the effectiveness of the proposed framework over several graph-based semi-supervised learning methods.

ℓ1-norm based safe semi-supervised learning.

In the past few years, Safe Semi-Supervised Learning (S3L) has received considerable attentions in machine learning field. Different researchers have proposed many S3L methods for safe exploitation of risky unlabeled samples which result in performance degradation of Semi-Supervised Learning (SSL). Nevertheless, there exist some shortcomings: (1) Risk degrees of the unlabeled samples are in advance defined by analyzing prediction differences between Supervised Learning (SL) and SSL; (2) Negative impacts of labeled samples on learning performance are not investigated. Therefore, it is essential to design a novel method to adaptively estimate importance and risk of both unlabeled and labeled samples. For this purpose, we present ℓ1-norm based S3L which can simultaneously reach the safe exploitation of the labeled and unlabeled samples in this paper. In order to solve the proposed ptimization problem, we utilize an effective iterative approach. In each iteration, one can adaptively estimate the weights of both labeled and unlabeled samples. The weights can reflect the importance or risk of the labeled and unlabeled samples. Hence, the negative effects of the labeled and unlabeled samples are expected to be reduced. Experimental performance on different datasets verifies that the proposed S3L method can obtain comparable performance with the existing SL, SSL and S3L methods and achieve the expected goal.

Deep Learning-Based Carotid Plaque Segmentation from B-Mode Ultrasound Images.

Carotid ultrasound measurement of total plaque area (TPA) provides a method for quantifying carotid plaque burden and monitoring changes in carotid atherosclerosis in response to medical treatment. Plaque boundary segmentation is required to generate the TPA measurement; however, training of observers and manual delineation are time consuming. Thus, our objective was to develop an automated plaque segmentation method to generate TPA from longitudinal carotid ultrasound images. In this study, a deep learning-based method, modified U-Net, was used to train the segmentation model and generate TPA measurement. A total of 510 plaques from 144 patients were used in our study, where the Monte Carlo cross-validation was used by randomly splitting the data set into 2/3 and 1/3 for training and testing. Two observers were trained to manually delineate the 510 plaques separately, which were used as the ground-truth references. Two U-Net models (M1 and M2) were trained using the two different ground-truth data sets from the two observers to evaluate the accuracy, variability and sensitivity on the ground-truth data sets used for training our method. The results of the algorithm segmentations of the two models yielded strong agreement with the two manual segmentations with the Pearson correlation coefficient r = 0.989 (p < 0.0001) and r = 0.987 (p < 0.0001). Comparison of the U-Net and manual segmentations resulted in mean TPA differences of 0.05 ± 7.13 mm2 (95% confidence interval: 14.02-13.02 mm2) and 0.8 ± 8.7 mm2 (17.85-16.25 mm2) for the two models, which are small compared with the TPA range in our data set from 4.7 to 312.8 mm2. Furthermore, the mean time to segment a plaque was only 8.3 ± 3.1 ms. The presented deep learning-based method described has sufficient accuracy with a short computation time and exhibits high agreement between the algorithm and manual TPA measurements, suggesting that the method could be used to measure TPA and to monitor the progression and regression of carotid atherosclerosis.

Improving motor imagery practice with synchronous action observation in stroke patients.

BACKGROUND: Action observation (AO) has the potential to improve motor imagery (MI) practice in stroke patients. However, currently only a few results are available on how to use AO effectively. OBJECTIVE: The aim of this study is to investigate whether MI practice can be improved more effectively by synchronous AO than by asynchronous AO. METHODS: Ten patients with upper limb motor dysfunction following stroke were selected as the participants. They were divided into two groups to perform MI practice combined with a daily conventional rehabilitation for four consecutive weeks. The control group was asked to perform MI guided by asynchronous AO (MIAAO), and the experimental group was asked to perform the same MI but guided by synchronous AO (MISAO). The event-related power decrease (ERD) in sensorimotor rhythms of electroencephalograph was calculated to reflect the sensorimotor cortex activation and to assess the cortex excitability during MI. Fugl-Meyer assessment (FMA) and pinch strength test (PST) were used to assess the limb motor recovery. RESULTS: The ERD pattern of the experimental group not only had greater amplitude and longer duration, but also included more frequency components. Furthermore, the effect sizes of ERD values between the two groups continuously increased (dES > 0.8) during the course of treatment. Moreover, the FMA and PST scores achieved with MISAO were also significantly higher than those achieved with MIAAO (p < 0.05). CONCLUSIONS: Compared with MIAAO, MISAO can enhance the excitation of sensorimotor cortex more effectively and lead to a more rapid neurorehabilitation of stroke patients.