An Estimation Method for Multi‐channel EEG Data Based on Canonical Correlation Analysis

Spectrum-Enhanced TRCA (SE-TRCA): A novel approach for direction detection in SSVEP-based BCI

This paper gives results from joint analyses of dual polarimety synthetic aperture radar data from the Sentinel-1 mission and optical data from the Sentinel-2 mission. The analyses are carried out by means of traditional canonical correlation analysis (CCA) and canonical information analysis (CIA). Where CCA is based on maximising correlation between linear combinations of the two data sets, CIA maximises mutual information between the two. CIA is a conceptually more pleasing method for the analysis of data with very different modalities such as radar and optical data. Although a little inconclusive as far as the change detection aspect is concerned, results show that CIA analysis gives conspicuously less noisy appearing images of canonical variates (CVs) than CCA. Also, the 2D histogram of the mutual information based leading CVs clearly reveals much more structure than the correlation based one. This gives promise for potentially better change detection results with CIA than can be obtained by means of CCA.

In recent years, the number of people with Alzheimer’s disease (AD) has grown worldwide. Since no cure has yet been found for the disease, detection and initiation of treatment is the best way to prevent the disease from progressing behavioral and physical symptoms in the patient. There are several methods to diagnose AD, and neuroimaging, as a noninvasive approach, reveals changes in the brain due to the disease. Diagnosis accuracy can be improved by fusion of various neuroimaging modalities. Canonical correlation analysis (CCA) and its extensions have been widely used for fusing multi-modal datasets, where healthy controls (HCs) are differentiated from patients by applying classification methods to canonical variables (CVs) resulting from CCA or its extensions. The goal of our study is to find an optimal method, from the perspective of accuracy and processing complexity, to diagnose patients with AD. This goal is achieved by fusing anatomical magnetic resonance imaging (MRI) and functional MRI (fMRI) data using CCA-based methods. Experimental results illustrate that HCs and AD patients can be classified using CVs obtained from structured and sparse CCA with greater than 90% accuracy.

The application of three multivariate analysis techniques (canonical discriminant analysis (CDA), principal component analysis (PCA), and canonical correlation analysis (CCA)) for evaluation of pasture botanical composition data is illustrated and discussed. CDA and PCA were used to describe differences in pasture botanical composition for different microsites within a pasture near Palmerston North, New Zealand. CCA could not be validly applied to this data set because a sampling strategy inappropriate for CCA had been used to collect the data. However, CCA is conceptually ideal for determining association between two groups of variables and CCA was used for a second data set from the Hawkes Bay region to establish association between differences in pasture botanical composition and differences in environmental variables. CCA identified a transition from white clover (Trifolium repens L.) to subterranean clover {Trifolium subterraneum L.) presence associated with decreasing rainfall, and a similar transition from ryegrass presence to browntop presence associated with altitude. There are practical difficulties in obtaining a suitable data set for canonical correlation, but with attention to sampling strategy a more precise definition of effect of environmental factors on pasture botanical composition would be expected.

This paper introduces a portable electroencephalogram (EEG) monitoring system for neonatal seizures. The hardware circuit can acquire 8 channels of EEG data or expand to 32 channels. It is characterized in high accuracy, small size, and low power consumption, which can meet the needs of long-term bedside monitoring of newborns. Users can choose different modes to view the historical data of the acquired EEG signals, or real-time data transmitted via WiFi. To further analyze and process the data, this paper introduces a machine learning-based scheme to automatically distinguish between normal and abnormal amplitude-integrated EEG (aEEG) signals. The program extracts the effective feature representation from the data and trains the model. The model can predict the class of unknown aEEG samples, which is helpful for further clinical detection of neonatal brain conditions.

Both canonical correlation analysis (CCA) and principal component analysis (PCA) were applied to atmospheric aerosol and trace gas concentrations and meteorological data collected in Chicago during the summer months of 2002, 2003, and 2004. Concentrations of ammonium, calcium, nitrate, sulfate, and oxalate particulate matter, as well as, meteorological parameters temperature, wind speed, wind direction, and humidity were subjected to CCA and PCA. Ozone and nitrogen oxide mixing ratios were also included in the data set. The purpose of statistical analysis was to determine the extent of existing linear relationship(s), or lack thereof, between meteorological parameters and pollutant concentrations in addition to reducing dimensionality of the original data to determine sources of pollutants. In CCA, the first three canonical variate pairs derived were statistically significant at the 0.05 level. Canonical correlation between the first canonical variate pair was 0.821, while correlations of the second and third canonical variate pairs were 0.562 and 0.461, respectively. The first canonical variate pair indicated that increasing temperatures resulted in high ozone mixing ratios, while the second canonical variate pair showed wind speed and humidity's influence on local ammonium concentrations. No new information was uncovered in the third variate pair. Canonical loadings were also interpreted for information regarding relationships between data sets. Four principal components (PCs), expressing 77.0 % of original data variance, were derived in PCA. Interpretation of PCs suggested significant production and/or transport of secondary aerosols in the region (PC1). Furthermore, photochemical production of ozone and wind speed's influence on pollutants were expressed (PC2) along with overall measure of local meteorology (PC3). In summary, CCA and PCA results combined were successful in uncovering linear relationships between meteorology and air pollutants in Chicago and aided in determining possible pollutant sources.

Canonical correlation analysis (CCA) is a powerful technique for assessing the relationship between two sets of variables. However, classical CCA does not distinguish between treatment and outcome variables, and it fails to account for potential confounding effects that may bias the estimated results. This study proposes a novel method called canonical causal analysis (Causal‐CCA) that extends CCA to estimate the canonical causal effects between multivariate continuous treatments and outcomes while adjusting for confounding variables. Causal effects are estimated within a framework where treatments and outcomes are prespecified. The proposed approach uses Gram–Schmidt orthogonalization and entropy balancing weights for multivariate treatments (EBMT) to extract unconfounded canonical causal variates and compute the canonical causal effects. Extensive simulations across various scenarios demonstrate that Causal‐CCA consistently outperforms classical CCA, weighted CCA with multivariate generalized propensity score weighting (MVGPS) and CCA with EBMT in terms of bias and mean squared error. The proposed method is applied to a real‐world dataset to study the causal relationship between psychological factors (self‐concept and motivation) and academic achievement (math and science scores), showing a significant positive causal effect of the psychological variables on academic performance.

Canonical correlation analysis is a statistical tool for investigating the presence of any patterns that simultaneously exist in two different data sets and compute the correlation between associated patterns. In this paper, dynamical systems for computing the canonical correlations and canonical variates between two data sets are derived. These systems are developed by optimizing constrained and unconstrained merit functions. To extract the actual individual canonical variates, some of these systems are weighted by a positive definite diagonal matrix, while others employed upper triangular matrices.

Objective:In this study; the relationships of several traits effects of Turkish Hair Goats were investigated by canonic correlation analysis. Therefore determining more effective factors on weaning weight of Hair goats are aimed.Material and Methods: The data were utilized from 1020 male and 1011 female Hair goat kids which were born in the period 2013 to 2016. The herd, year of birth, birth type and sex are analyzed as the first set of variables (X); weight at birth, weight at thirty day of age and weight at weaning age effects are analyzed as the second set of variables (Y). By performing canonical correlation analysis, canonical correlations between the first, second and third pair of canonical variates were estimated and three canonical correlation were found significant (P < 0.01).Results:The biggest correlations between year of birth, sex, birth type variables and v1, v2 and v3 canonical variates were found -0.77%, 0.87% and -0.82%, respectively. The biggest correlations between weight at birth, weight at thirty day of age and weight at weaning age and w1, w2 and w3 canonical variates were found 0.84%, -0.97% and -0.78%, respectively. As a result it has been determined that sex and birth type play an important role in the formation of the goat kid’s body weights.Conclusion:Hair goats reared in West Anatolia are slaughtered mostly at the end of the weaning. Because of this, assessment the factors effect on weaning weight are important. Canonical correlation analysis therefore is used to analyze all factors at the same time to define relationships between each other of weaning traits for hair goats.

The objective of this study was to investigate the multivariate relationship among body protein, fat, and macrominerals in Saanen goats of different sexes (castrated males, females, and intact males) using canonical correlation analysis. Individual records of 274 Saanen goats combined from 10 studies was used. Two sets of body constituents were established: the first one contained variables related to body protein or fat (canonical variate U) and the second contained variables related to body calcium, phosphorus, magnesium, sodium, or potassium (canonical variate V). Two canonical pairs were identified for each sex. However, irrespective of sex, first canonical pairs accounted for more than 87% of variance of the dependent variables, these only being used for the analysis. For canonical variate U1, canonical weights for protein were greater than that for fat (in castrated males, protein = 0.62 and fat = 0.41; in females, protein = 0.96 and fat = 0.039; and in intact males, protein = 0.81 and fat = 0.20). For canonical variate V1, in males, the highest canonical weights were for potassium and phosphorus, whereas the lowest were for calcium (in castrated males, potassium = 0.485 > phosphorus = 0.259 > magnesium = 0.206 > sodium = 0.129 > calcium = 0.081, and in intact males, potassium = 0.499 > phosphorus = 0.459 > sodium = 0.105 > magnesium = 0.024 > calcium = 0.001). On the other hand, in females, the highest canonical weights were for potassium and calcium, whereas the lowest was for magnesium (potassium = 0.539 > calcium = 0.201 > phosphorus = 0.178 > sodium = 0.088 > magnesium = 0.081). The current results may help to understand the role of sex on strength and nature of the association among body protein, fat, and macrominerals in growing Saanen goats.

Canonical correlation analysis of marine macrobenthos survey data

Intensive land use is requiring more detailed information about patterns and magnitudes of soil variability than can be acquired through traditional soil survey techniques. Discriminant analysis is a mathematical method of numerical classification which could be used to identify discrete populations of soils in their natural settings. The hypothesis of this study was that discriminant analysis could be used to group soils on landtypes on the Mid-Cumberland Plateau. A large data set (132 observations of 29 soil variables) was collected from three landtypes at two Cumberland Plateau locations. Discriminant analysis was used to classify the soil observations into landtypes. Canonical correlation was used to identify soil properties most responsible for separating soils into groups related to landtypes. Not all of the collected soil properties were important discriminators, so variables with low canonical loading scores were eliminated. A total of 13 soil variables representing three genetic soil horizons was required to correctly classify all 132 observations into correct landtypes. Canonical correlations were 0.979 and 0.970 with 29 variables and 0.968 and 0.941 with 13 variables on canonical variates one and two, respectively. Soil variables from Bt horizons alone did not classify all observations into correct landtypes. Discriminant analysis, in conjunction with canonical correlation, shows promise for identifying key variables for numerically classifying soils into related populations.

Brain-computer interface (BCI) is used for non-muscular communication between the human brain and an external device. One of the most commonly utilized phenomena in BCI are steady-state visually evoked potentials (SSVEP) that are measured with electroencephalography (EEG). If the subject focuses his/her attention on the stimulus flashing with specified frequency on the computer screen, a signal fluctuations with the same frequency may be observed in his or her visual cortex. Canonical Correlation Analysis (CCA) was adapted to EEG signal analysis and the outcomes of this straightforward approach are more promising in terms of obtaining decoding accuracy than conventional methods. In CCA two sets of data are being compared: a multichannel signal, and a reference prior corresponding to the classes between which the distinction is being made. Category of the signal (e.g., 8 Hz vs 14 Hz) is determined on the basis of the correlation coefficient between the acquired signal and several reference signals. The aim of this paper is to present an idea of utilizing genetic algorithms to generate these reference priors. Outcomes of our analyses suggest that a genetic approach (with signal encoding in the form of the binary code) and tournament selection method may improve the accuracy of SSVEP classification. Further work may help to improve the accuracy even more, with different encoding scheme or alternative and/or more sophisticated selection approach.

Purpose: To evaluate the impact of complexity indices on the plan quality and deliverability of volumetric modulated arc therapy (VMAT), and to determine the most significant parameters in the generation of an ideal VMAT plan. Methods: A multi-dimensional exploratory statistical method, canonical correlation analysis (CCA) was adopted to study the correlations between VMAT parameters of complexity, quality and deliverability, as well as their contribution weights with 32 two-arc VMAT nasopharyngeal cancer (NPC) patients and 31 one-arc VMAT prostate cancer patients. Results: The MU per arc (MU/Arc) and MU per control point (MU/CP) of NPC were 337.8±25.2 and 3.7±0.3, respectively, which were significantly lower than those of prostate cancer patients (MU/Arc : 506.9±95.4, MU/CP : 5.6±1.1). The plan complexity indices indicated that two-arc VMAT plans were more complex than one-arc VMAT plans. Plan quality comparison confirmed that one-arc VMAT plans had a high quality than two-arc VMAT plans. CCA results implied that plan complexity parameters were highly correlated with plan quality with the first two canonical correlations of 0.96, 0.88 (both p<0.001) and significantly correlated with deliverability with the first canonical correlation of 0.79 (p<0.001), plan quality and deliverability was also correlated with the first canonical correlation of 0.71 (p=0.02). Complexity parameters of MU/CP, segment area (SA) per CP, percent of MU/CP less 3 and planning target volume (PTV) were weighted heavily in correlation with plan quality and deliveability . Similar results obtained from individual NPC and prostate CCA analysis. Conclusion: Relationship between complexity, quality, and deliverability parameters were investigated with CCA. MU, SA related parameters and PTV volume were found to have strong effect on the plan quality and deliverability. The presented correlation among different quantified parameters could be used to improve the plan quality and the efficiency of the radiotherapy process when creating a complex VMAT plan.

ObjectiveIn recent years, with the development of machine learning and deep learning technologies, an increasing number of research works have begun using these technologies for automatic seizure detection in EEG signals. However, existing automatic seizure detection algorithms primarily focus on the features of individual EEG channels and pay less attention to the inter-channel relationships. This results in insufficient extraction of spatiotemporal information from multi-channel EEG data, affecting the final seizure detection performance.MethodsTherefore, this paper proposes an automatic seizure detection method based on the combination of Graph Attention Networks (GAT) and Transformer networks. Specifically, GAT is used as the front end for extracting spatial features, fully leveraging the topological structure of different EEG channels. Meanwhile, the Transformer network is used as the back end to explore temporal relationships and make final decisions based on the states before and after the current moment.ResultsExperiments were conducted on the CHB-MIT and TUH datasets with ten-fold cross-validation. The final seizure detection accuracies on the two datasets were 98.62 and 98.12%, respectively, with the model’s performance surpassing or being comparable to current state-of-the-art models.ConclusionThe proposed hybrid algorithm combines the advantages of two deep learning models, fully exploring the spatiotemporal correlations between EEG channels. Experiments on public datasets demonstrate the effectiveness of this method, significantly advancing the development of automatic seizure detection.