Chih-I Hung

Recognition of motor imagery electroencephalography using independent component analysis and machine classifiers.

Motor imagery electroencephalography (EEG), which embodies cortical potentials during mental simulation of left or right finger lifting tasks, can be used to provide neural input signals to activate a brain computer interface (BCI). The effectiveness of such an EEG-based BCI system relies on two indispensable components: distinguishable patterns of brain signals and accurate classifiers. This work aims to extract two reliable neural features, termed contralateral and ipsilateral rebound maps, by removing artifacts from motor imagery EEG based on independent component analysis (ICA), and to employ four classifiers to investigate the efficacy of rebound maps. Results demonstrate that, with the use of ICA, recognition rates for four classifiers (fisher linear discriminant (FLD), back-propagation neural network (BP-NN), radial-basis function neural network (RBF-NN), and support vector machine (SVM)) improved significantly, from 54%, 54%, 57% and 55% to 70.5%, 75.5%, 76.5% and 77.3%, respectively. In addition, the areas under the receiver operating characteristics (ROC) curve, which assess the quality of classification over a wide range of misclassification costs, also improved from .65, .60, .62, and .64 to .74, .76, .80 and .81, respectively.

Morphological regionalization using fetal magnetic resonance images of normal developing brains

Regional differences in human brain development during infancy have been studied for many years, but little is known about how regionalization of the brain proceeds during intrauterine life. We investigated the regionalization of cerebral volume and cortical convolutions based on the volumetric magnetic resonance images (MRIs) of 43 fetuses, ranging from 21 to 37 weeks of gestation. Two plausible parcellations of MRI are proposed, and curvature index together with gyrification index are used to quantify the regional cortical convolutions. Our results elucidate that the cortical foldings among different brain regions develop at comparable rates, suggesting a similar uniformity of changes in size of the cortical sheet in these regions over time. On the contrary, the growth of the cerebral volume presents regional difference, with the frontal and parieto‐temporal regions growing significantly faster than other regions due to the contribution from expansion of basal ganglia. This quantitative regional information suggests that cerebral volume is not a relevant parameter to measure in relation to gyrification, and that the size of the cortical sheet is more likely to be directly related to cortical folding. The availability of quantitative regional information on normal fetal brains in utero will allow clinical application of this information when probing neurodevelopmental disorders in the future.

Early detection of periodic sharp wave complexes on EEG by independent component analysis in patients with Creutzfeldt-Jakob disease.

Summary: Sporadic Creutzfeldt-Jakob disease (sCJD) is the most common human prion disease. EEG is the method of choice to support the diagnosis of a human prion disease. Periodic sharp wave complexes (PSWCs) on the EEG usually indicate a progressive stage of CJD. However, PSWCs only become obvious at around 8 to 12 weeks after the onset of clinical symptoms, and in a few cases, even later. Independent component analysis (ICA) is a new technique to separate statistically independent components from a mixture of data. This study recruited seven patients who fit the criteria of CJD between 2002 and 2005 and 10 patients with Alzheimer’s disease (AD) as control subjects. Using an ICA algorithm, we were able to split typical PSWCs into several independent temporal components in conjunction with spatial maps. The PSWCs were not observed in the initial EEG studies of patients with either AD or CJD. However, the ICA algorithm was able to extract periodic discharges and epileptiform discharges from raw EEG of patients with CJD at as early as 3 to 5 weeks after disease onset. Such discharges otherwise could hardly be discerned by visual inspection. In conclusion, ICA may increase the sensitivity of EEG and facilitate the early diagnosis of CJD.

Reorganization of functional connectivity during the motor task using EEG time–frequency cross mutual information analysis

OBJECTIVE This study investigates the functional organization of cortical networks during self-determinant arm movement using the time sequences of the alpha (8-12 Hz) and beta (16-25 Hz) bands. METHODS The time-frequency cross mutual information (TFCMI) method was used to estimate the EEG functional connectivity in the alpha and beta bands for seven healthy subjects during four functional states: the resting, preparing, movement-onset, and movement-offset states. RESULTS In the preparing state, the maintenance of the central-executive network (CEN, prefrontal-parietal connection) suppressed the motor network in the alpha band to plan the next movement, whereas the CEN was deactivated in the beta band to retain visual attention (the frontal-occipital connection). A significant decrease of the CEN in the alpha band occurred after a visual cue in the movement-onset state, followed by a significant increase in motor-network connectivity in the beta band until the movement-offset state. CONCLUSIONS The temporal-spectral modulation mechanism allows the brain to manifest multiple functions subject to energy budget. SIGNIFICANCE The TFCMI method was employed to estimate EEG functional connectivity and effectively demonstrate the reorganization process between four functional states.

Blind Source Separation of Concurrent Disease-Related Patterns from EEG in Creutzfeldt–Jakob Disease for Assisting Early Diagnosis

Creutzfeldt–Jakob disease (CJD) is a rare, transmissible and fatal prion disorder of brain. Typical electroencephalography (EEG) patterns, such as the periodic sharp wave complexes (PSWCs), do not clearly emerge until the middle stage of CJD. To reduce transmission risks and avoid unnecessary treatments, the recognition of the hidden PSWCs forerunners from the contaminated EEG signals in the early stage is imperative. In this study, independent component analysis (ICA) was employed on the raw EEG signals recorded at the first admissions of five patients to segregate the co-occurrence of multiple disease-related features, which were difficult to be detected from the smeared EEG. Clear CJD-related waveforms, i.e., frontal intermittent rhythmical delta activity (FIRDA), fore PSWCs (triphasic waves) and periodic lateralized epileptiform discharges (PLEDs), have been successfully and simultaneously resolved from all patients. The ICA results elucidate the concurrent appearance of FIRDA and PLEDs or triphasic waves within the same EEG epoch, which has not been reported in the previous literature. Results show that ICA is an objective and effective means to extract the disease-related patterns for facilitating the early diagnosis of CJD.

Classification of bipolar disorder using basal-ganglia-related functional connectivity in the resting state

The emotional and cognitive symptoms of bipolar disorder (BD) are suggested to involve in a distributed neural network. The resting-state functional magnetic resonance imaging (fMRI) offers an important tool to investigate the alterations in brain network level of BD. The aim of this study was to discriminate BD patients from healthy controls using whole-brain resting-state functional connectivity patterns. The majority of most discriminating functional connectivities were between the basal ganglia and three core neurocognitive networks, including the default mode, executive control and salience networks. Using these resting-state functional connectivities between the basal ganglia and three core neurocognitive networks as the features, the clustering accuracy achieved 90%.

Regional quantification of developing human cortical shape with a three-dimensional surface-based magnetic resonance imaging analysis in utero.

Although regional differences in cerebral volume have been revealed in developing human brains, little is known regarding the regionalization of cortical shape. This study documented the regional and quantitative shape difference of cortical surfaces for in utero normal fetal brains over a time period essential for the formation of primary cortical folding (22–33 weeks). Each brain surface with complete three‐dimensional morphology was manually extracted from the reconstructed image, which combined surface information from three orthogonal magnetic resonance images in utero. An innovative parcellation was used to dissect the fetal brains into frontal, parietal, temporal and occipital lobes, and to avoid the determination of non‐existent and immature sulci for young fetuses. Distinct cortical shapes were encoded by the shape index automatically. The results of this study show faster shape changes in the occipital lobe than in other regions. Both regional and global shape patterns show that the gyral surface smoothens, whereas the sulcal surface becomes more angular, with gestational age. In addition, the smoothing of gyri is related mainly to the changes in shape of gyral crowns. This study presents the regional differences in early gyrification from the novel aspect of shape. The results of shape pattern analysis for normal fetuses may act as a reference in assessments of prenatal brain pathology and in extensive comparisons between various life stages.

Cortical Shape and Curvedness Analysis of Structural Deficits in Remitting and Non-Remitting Depression

In morphometric neuroimaging studies, the relationship between brain structural changes and the antidepressant treatment response in patients with major depressive disorder has been explored to search depression-trait biomarkers. Although patients were treated with serotonin-related drugs, whether the same treatment resulted in remission and non-remission in depressed patients is currently under investigation. We recruited 25 depressed patients and 25 healthy controls and acquired volumetric magnetic resonance imaging of each participant. We used the shape index and curvedness to classify cortical shapes and quantify shape complexities, respectively, in studying the pharmacological effect on brain morphology. The results showed that different regions of structural abnormalities emerged between remitting and non-remitting patients when contrasted with healthy controls. In addition to comparing structural metrics in each cortical parcellation, similar to the traditional voxel-based morphometric method, we highlighted the importance of structural integrity along the serotonin pathway in response to medication treatment. We discovered that disrupted serotonin-related cortical regions might cause non-remission to antidepressant treatment from a pharmacological perspective. The anomalous areas manifested in non-remitting patients were mainly in the frontolimbic areas, which can be used to differentiate remitting from non-remitting participants before medication treatment. Because non-remission is the failure to respond to treatment with serotonin-related drugs, our method may help clinicians choose appropriate medications for non-remitting patients.

Enhancement of Signal-to-noise Ratio of Peroneal Nerve Somatosensory Evoked Potential Using Independent Component Analysis and Time-Frequency Template

This study aims to recover the somatosensory evoked potentials (SSEPs) from the smearing electroencephalography (EEG) recordings using independent component analysis (ICA) in conjunction with the proposed timefrequency SSEP template (TF-SSEP). The SSEPs induced from patients with the impaired motor functions exhibit longer latency and lower amplitude than the normal SSEPs and are inevitably contaminated by artifacts and environmental noise. Although ICA has been demonstrated as a novel technique to segregate the EEG into independent sources, the selection of task-related components needs to be further elaborated. The TF-SSEP template, generated by the Morelet wavelet transformation of the averaged SSEPs from three normal subjects, was used to automatically extract the SSEP-related features. The performance of the TF-SSEP template was further validated using EEGs through the left and right peroneal nerve stimulation of four stroke patients. After ICA decomposition, the sources were selected for reconstruction if their correlation coefficients with the TF-SSEP template were higher than the predetermined threshold. On the other hand, the unselected sources were considered as the event-unrelated components or artifacts. Among all patients, the topography maps at four peak times, namely P40, N45, P60 and N75, showed higher contrast in the vicinity of the foot-associated motor area, and the resolved SSEPs demonstrated uncontaminated waveforms in comparison with the conventionally averaging method. This indicated that the proposed method can remarkably suppress artifacts and effectively extracted the SSEP-related features.