The stable decoding of movement parameters using neural activity is crucial for the success of brain-machine interfaces (BMIs). However, neural activity can be unstable over time, leading to changes in the parameters used for decoding movement, which can hinder accurate movement decoding. To tackle this issue, one approach is to transfer neural activity to a stable, low-dimensional manifold using dimensionality reduction techniques and align manifolds across sessions by maximizing correlations of the manifolds. However, the practical use of manifold stabilization techniques requires knowledge of the true subject intentions such as target direction or behavioral state. To overcome this limitation, an automatic unsupervised algorithm is proposed that determines movement target intention before manifold alignment in the presence of manifold rotation and scaling across sessions. This unsupervised algorithm is combined with a dimensionality reduction and alignment method to overcome decoder instabilities. The effectiveness of the BMI stabilizer method is represented by decoding the two-dimensional (2D) hand velocity of two rhesus macaque monkeys during a center-out-reaching movement task. The performance of the proposed method is evaluated using correlation coefficient and R-squared measures, demonstrating higher decoding performance compared to a state-of-the-art unsupervised BMI stabilizer. The results offer benefits for the automatic determination of movement intents in long-term BMI decoding. Overall, the proposed method offers a promising automatic solution for achieving stable and accurate movement decoding in BMI applications.

MOTIVATION: The aim of quantitative structure-activity prediction (QSAR) studies is to identify novel drug-like molecules that can be suggested as lead compounds by means of two approaches, which are discussed in this article. First, to identify appropriate molecular descriptors by focusing on one feature-selection algorithms; and second to predict the biological activities of designed compounds. Recent studies have shown increased interest in the prediction of a huge number of molecules, known as Big Data, using deep learning models. However, despite all these efforts to solve critical challenges in QSAR models, such as over-fitting, massive processing procedures, is major shortcomings of deep learning models. Hence, finding the most effective molecular descriptors in the shortest possible time is an ongoing task. One of the successful methods to speed up the extraction of the best features from big datasets is the use of least absolute shrinkage and selection operator (LASSO). This algorithm is a regression model that selects a subset of molecular descriptors with the aim of enhancing prediction accuracy and interpretability because of removing inappropriate and irrelevant features. RESULTS: To implement and test our proposed model, a random forest was built to predict the molecular activities of Kaggle competition compounds. Finally, the prediction results and computation time of the suggested model were compared with the other well-known algorithms, i.e. Boruta-random forest, deep random forest and deep belief network model. The results revealed that improving output correlation through LASSO-random forest leads to appreciably reduced implementation time and model complexity, while maintaining accuracy of the predictions. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Abstract Cardiac failure, such as myocardial infarction (MI), is one of the most serious causes of mortality worldwide. MI is the sign of cardiac cell damage as a result of decreased blood oxygen level, which causes some morphological changes in the form of 12‐lead electrocardiogram (ECG) waves including T‐wave, Q‐wave, and ST‐segment. The main goal of this study is to represent vectorcardiography (VCG) as a complementary diagnostic tool of the ECG method to discriminate the various type of MI from normal cases. The system proposed in this study was analysed on the Physikalisch‐Technische Bundesanstalt diagnostic ECG database and a recorded signal database for 80 MI and 52 healthy cases. Each record consists of 15 ECG and VCG signals. In this study, tridimensional morphological features were applied to the classification and regression tree (CART) and the feedforward neural network classifier. To classify MI cases from healthy control cases of our recorded database, classification and regression tree achieved the same results when VCG features or ECG features were applied with an accuracy of 99.4%, a sensitivity of 100%, and a specificity of 98.7%. Further, by using VCG Octant features with this current method, anterior‐MI and inferior‐MI were separated with an accuracy of 98.9%, a sensitivity of 98%, and a specificity of 100%. The outcomes prove that the VCG features performed more accurately than ECG features in MI localisation.

This paper proposes a mixed low-rank approximation and second-order tensor-based total variation (LRSOTTV) approach for the super-resolution and denoising of retinal optical coherence tomography (OCT) images through effective utilization of nonlocal spatial correlations and local smoothness properties. OCT imaging relies on interferometry, which explains why OCT images suffer from a high level of noise. In addition, data subsampling is conducted during OCT A-scan and B-scan acquisition. Therefore, using effective super-resolution algorithms is necessary for reconstructing high-resolution clean OCT images. In this paper, a low-rank regularization approach is proposed for exploiting nonlocal self-similarity prior to OCT image reconstruction. To benefit from the advantages of the redundancy of multi-slice OCT data, we construct a third-order tensor by extracting the nonlocal similar three-dimensional blocks and grouping them by applying the k-nearest-neighbor method. Next, the nuclear norm is used as a regularization term to shrink the singular values of the constructed tensor in the non-local correlation direction. Further, the regularization approaches of the first-order tensor-based total variation (FOTTV) and SOTTV are proposed for better preservation of retinal layers and suppression of artifacts in OCT images. The alternative direction method of multipliers (ADMM) technique is then used to solve the resulting optimization problem. Our experiments show that integrating SOTTV instead of FOTTV into a low-rank approximation model can achieve noticeably improved results. Our experimental results on the denoising and super-resolution of OCT images demonstrate that the proposed model can provide images whose numerical and visual qualities are higher than those obtained by using state-of-the-art methods.

In this paper, a new statistical model is proposed for the single image super-resolution of retinal Optical Coherence Tomography (OCT) images. OCT imaging relies on interfero-metry, which explains why OCT images suffer from a high level of noise. Moreover, data subsampling is carried out during the acquisition of OCT A-scans and B-scans. So, it is necessary to utilize effective super-resolution algorithms to reconstruct high-resolution clean OCT images. In this paper, a nonlocal sparse model-based Bayesian framework is proposed for OCT restoration. For this reason, by characterizing nonlocal patches with similar structures, known as a group, the sparse coefficients of each group of OCT images are modeled by the scale mixture models. In this base, the coefficient vector is decomposed into the point-wise product of a random vector and a positive scaling variable. Estimation of the sparse coefficients depends on the proposed distribution for the random vector and scaling variable where the Laplacian random vector and Generalized Extreme-Value (GEV) scale parameter (Laplacian+GEV model) show the best goodness of fit for each group of OCT images. Finally, a new OCT super-resolution method based on this new scale mixture model is introduced, where the maximum a posterior estimation of both sparse coefficients and scaling variables are calculated efficiently by applying an alternating minimization method. Our experimental results prove that the proposed OCT super-resolution method based on the Laplacian+GEV model outperforms other competing methods in terms of both subjective and objective visual qualities.

Computer-aided diagnosis (CAD) of retinal pathologies is a current active area in medical image analysis. Due to the increasing use of retinal optical coherence tomography (OCT) imaging technique, a CAD system in retinal OCT is essential to assist ophthalmologist in the early detection of ocular diseases and treatment monitoring. This paper presents a novel CAD system based on a multi-scale convolutional mixture of expert (MCME) ensemble model to identify normal retina, and two common types of macular pathologies, namely, dry age-related macular degeneration, and diabetic macular edema. The proposed MCME modular model is a data-driven neural structure, which employs a new cost function for discriminative and fast learning of image features by applying convolutional neural networks on multiple-scale sub-images. MCME maximizes the likelihood function of the training data set and ground truth by considering a mixture model, which tries also to model the joint interaction between individual experts by using a correlated multivariate component for each expert module instead of only modeling the marginal distributions by independent Gaussian components. Two different macular OCT data sets from Heidelberg devices were considered for the evaluation of the method, i.e., a local data set of OCT images of 148 subjects and a public data set of 45 OCT acquisitions. For comparison purpose, we performed a wide range of classification measures to compare the results with the best configurations of the MCME method. With the MCME model of four scale-dependent experts, the precision rate of 98.86%, and the area under the receiver operating characteristic curve (AUC) of 0.9985 were obtained on average.

Symmetry analysis of right and left eyes can be a useful tool for early detection of eye diseases. In this study, we want to compare the Optical Coherent Tomography (OCT) images captured from optic nerve head (ONH) of right and left eyes. To do this, it is necessary to align the OCT data and compare equivalent B-scans in right and left eyes. For this reason, since the fovea-ONH axes in OCT data are not available due to small field of view in OCT, at first the projection of OCT data of each eye is registered to its corresponding fundus image using extracted vessels by Hessian analysis of directional curvelet subbands. Then, by alignment of fundus images of right and left eyes according to their automatically detected fovea-ONH axes, OCT projections are also aligned. After alignment of OCT projections, aligned B-scans are estimated and used for comparing different parameters such as cup-to-disk ratio (CDR). Using aligned B-scans, two signals of CDRs are obtained from two eyes which each point in these signals corresponds to CDR in a specific part of ONH, i.e., a point-to-point comparison between CDRs of right and left eyes is provided which has potential to lead to a new imaging biomarker for eye disease detection.

Optic Disc (OD) detection in retinal fundus images is a crucial stage for the automation of a screening system in diabetic ophthalmology. Most researches for automatic localization of OD benefit the regions of vessels. In this paper, we present a fast and novel method based on the Circlet Transform to detect OD in digital retinal fundus images that doesn't utilize the location of the vessels. First, each R, G and B band is enhanced using CLAHE method. Then, the enhanced image in RGB color space is converted to L*a*b one. Next, the Circlet transform is applied to the L* band, and finally, the Circlet transform coefficients are analyzed to find the location of the OD. The proposed algorithm is implemented on DRIVE dataset and the experimental results show a very well OD localization. The correct rate of the proposed method is 95% even though it doesn't utilize the vessels' structure.

Asymmetry analysis is a challenging step in computerized early diagnosis of Diabetic retinopathy (DR) which provides an opportunity for early treatment. In this study to compare the patterns of vascular in right and left eyes, a combination of fractal analysis and radon transformation is investigated to provide both statistical distribution of the vessel thickness, and their geometrical distribution. For this purpose, the vessel segmentation and skeletonizetion are performed and the vessels' thickness map (VTM) is obtained. Then, the fractal dimension (FD) is found on various versions like the segmented vessels, skeletonized vessels, VTM, and radon transform (RT) of VTM in right and left eyes for asymmetry analysis. According to the obtained results for mean/SD values of the differences of FDs in right and left eyes and p-values, we conclude that RT of VTM is able to better discriminate two eyes from each other and accordingly, it can be used as a powerful feature for comparison of the symmetry/asymmetry in fundus images. Our evaluation results show that a difference of 0.33 ± 0.11 between FD of VTM's RT in left and right eyes is expected for normal subjects.

Leukemia (a cancer of leukocytes) basically develops in the bone marrow. Acute myelogenous leukemia (a type of leukemia) has eight sub-types according to French-American-British classification. These forms can be visually observed by pathologists using microscopic images of infected cells. However, identification task is tedious and usually difficult due to varying features. Automatic leukemia detection is an important topic in the domain of cancer diagnosis. This paper presents a novel method based on dictionary learning and sparse representation for detecting and classification of different sub-types of AML. For each class, two intensity and label dictionaries are designed for representation using image patches of training samples. New image is represented by all dictionaries and the one with minimum error determine the type of class. We considered M2, M3 and M5 sub-types for evaluation of the method. The initial implementing of the proposed method achieved 97.53% average accuracy for different sub-types of AML.

In this experiment, a gene selection technique was proposed to select a robust gene signature from microarray data for prediction of breast cancer recurrence. In this regard, a hybrid scoring criterion was designed as linear combinations of the scores that were determined in the mutual information (MI) domain and protein-protein interactions network. Whereas, the MI-based score represents the complementary information between the selected genes for outcome prediction; and the number of connections in the PPI network between the selected genes builds the PPI-based score. All genes were scored by using the proposed function in a hybrid forward-backward gene-set selection process to select the optimum biomarker-set from the gene expression microarray data. The accuracy and stability of the finally selected biomarkers were evaluated by using five-fold cross-validation (CV) to classify available data on breast cancer patients into two cohorts of poor and good prognosis. The results showed an appealing improvement in the cross-dataset accuracy in comparison with similar studies whenever we applied a primary signature, which was selected from one dataset, to predict survival in other independent datasets. Moreover, the proposed method demonstrated 58-92 percent overlap between 50-genes signatures, which were selected from seven independent datasets individually.

The ability to transform orthogonal 3-lead Vector-cardiogram (VCG) to 12-lead Electrocardiogram (ECG) enables the use of fewer leads for computer visualization, signal analysis and wireless transmission of signals. This can also improve mobility, albeit limited, to the patients. We presented a least square (LS)-based approach to transform 3-lead Frank VCG to 12-lead ECG signals and vice versa, using partial linear transformation. Also our partial linear transformation function would be compared with Dower and affine transformation functions. The VCG and ECG signals of 40 healthy persons are acquired in this study. The results show that for healthy subjects, our partial linear LS method that is maps 3-lead VCG to12-lead ECG more accurately than both Dower and affine transformations of the ECG recordings.

Various techniques are used in diagnosing cardiac diseases. One of these techniques is using electrocardiogram (ECG) tool. In special cardiac cases like atrial fibrillation and posterior myocardial infraction the cardiologist need some information from posterior side of the patient heart, that it can be achieved by using right-posterior ECG method (17 lead ECG). In right-posterior method, position of the patient must be changed in his/her side, so time is waste and patient would be more tired because of taking ECG signals two times. In this study vectorcardiogram (VCG) signals are used as a tool for providing posterior information of the heart. However because for cardiologists is much easier to work with ECG signals for detecting some cardiac diseases, in this study a new method using partial linear transformation is introduced to get posterior ECG leads (V7, V8, V9) from VCG signal. VCG and ECG signals that were used in this study obtained from 30 healthy persons. We presented a statistical approach to transform 3-lead Frank VCG to 15-lead ECG signals and vice versa, based on partial linear transformation (Least Square Method). Also our linear transformation function would be compared with affine transformation functions. The recorder device was Cardiax digital recorder system. The results show that for healthy subjects, the partial linear transformation (least square method) that is presented in this paper maps 3-lead VCG to15-lead ECG, is more accurate than affine transformation function. Regarding the obtained results in this study, ECG signals that derived from VCG signals by using our method was more similar to measured ECG signals than ones derived by using affine transformation. Therefore, by using this transformation function achieving to posterior information of the case heart would be more accurate and useful.

Red lesions in the form of Microaneurysms (MAs) and Hemorrhages (HMs) are among the first explicit signs of diabetic retinopathy (DR). Hence robust detection of these lesions is an important diagnostic task in computer assistance systems. In this paper we present a new curvelet based algorithm to separate these red lesions from the rest of the color retinal image. In order to prevent fovea to be considered as red lesion, we introduce a new illumination equalization algorithm and apply that to green plane of retinal image. In the next stage, we apply digital curvelet transform (DCUT) to produced enhanced image and modify curvelet coefficients in order to lead red objects to zero. Then we separate these lesions as candidate region by applying an appropriate threshold. Finally, the total structure of blood vessel is extracted employing a curvelet-based technique and the false positives (FPs) are eliminated by subtracting the vessel structure from the candidate images. Experiments on 89 retinal images of diabetic patients indicate that we are able to achieve 94% sensitivity and 87% specificity in detection of red lesion.

This paper presents an efficient method for automatic extraction of blood vessels in retinal images to improve the detection of low contrast and narrow vessels. The proposed algorithm is composed of four steps: curvelet-based contrast enhancement, match filtering, curvelet-based edge extraction, and length filtering. In this base, after reconstruction of enhanced image from the modified curvelet coefficients, match filtering is used to intensify the blood vessels. Then we employ curvelet transform to segment vessels from its background and finally the length filtering is used to remove the misclassified pixels. The performance of algorithm is evaluated on DRIVE [1] databases and compared with those obtained from a hand-labeled ground truth. Since the curvelet transform is well-suited to handle curve discontinuities, we achieve an area under ROC curve of 0.9631 that demonstrates improved performance of proposed algorithm compared with known techniques.