A Convolutional Neural Network for SSVEP Identification by Using a Few-Channel EEG.

The application of wearable electroencephalogram (EEG) devices is growing in brain-computer interfaces (BCI) owing to their good wearability and portability. Compared with conventional devices, wearable devices typically support fewer EEG channels. Devices with few-channel EEGs have been proven to be available for steady-state visual evoked potential (SSVEP)-based BCI. However, fewer-channel EEGs can cause the BCI performance to decrease. To address this issue, an attention-based complex spectrum-convolutional neural network (atten-CCNN) is proposed in this study, which combines a CNN with a squeeze-and-excitation block and uses the spectrum of the EEG signal as the input. The proposed model was assessed on a wearable 40-class dataset and a public 12-class dataset under subject-independent and subject-dependent conditions. The results show that whether using a three-channel EEG or single-channel EEG for SSVEP identification, atten-CCNN outperformed the baseline models, indicating that the new model can effectively enhance the performance of SSVEP-BCI with few-channel EEGs. Therefore, this SSVEP identification algorithm based on a few-channel EEG is particularly suitable for use with wearable EEG devices.

Alternative muscle synergy patterns of upper limb amputees.

Myoelectric hand prostheses are effective tools for upper limb amputees to regain hand functions. Much progress has been made with pattern recognition algorithms to recognize surface electromyography (sEMG) patterns, but few attentions was placed on the amputees' motor learning process. Many potential myoelectric prostheses users could not fully master the control or had declined performance over time. It is possible that learning to produce distinct and consistent muscle activation patterns with the residual limb could help amputees better control the myoelectric prosthesis. In this study, we observed longitudinal effect of motor skill learning with 2 amputees who have developed alternative muscle activation patterns in response to the same set of target prosthetic actions. During a 10-week program, amputee participants were trained to produce distinct and constant muscle activations with visual feedback of live sEMG and without interaction with prosthesis. At the end, their sEMG patterns were different from each other and from non-amputee control groups. For certain intended hand motion, gradually reducing root mean square (RMS) variance was observed. The learning effect was also assessed with a CNN-LSTM mixture classifier designed for mobile sEMG pattern recognition. The classification accuracy had a rising trend over time, implicating potential performance improvement of myoelectric prosthesis control. A follow-up session took place 6 months after the program and showed lasting effect of the motor skill learning in terms of sEMG pattern classification accuracy. The results indicated that with proper feedback training, amputees could learn unique muscle activation patterns that allow them to trigger intended prosthesis functions, and the original motor control scheme is updated. The effect of such motor skill learning could help to improve myoelectric prosthetic control performance.

Deep Learning-Based Auto-Segmentation of Spinal Cord Internal Structure of Diffusion Tensor Imaging in Cervical Spondylotic Myelopathy.

Cervical spondylotic myelopathy (CSM) is a chronic disorder of the spinal cord. ROI-based features on diffusion tensor imaging (DTI) provide additional information about spinal cord status, which would benefit the diagnosis and prognosis of CSM. However, the manual extraction of the DTI-related features on multiple ROIs is time-consuming and laborious. In total, 1159 slices at cervical levels from 89 CSM patients were analyzed, and corresponding fractional anisotropy (FA) maps were calculated. Eight ROIs were drawn, covering both sides of lateral, dorsal, ventral, and gray matter. The UNet model was trained with the proposed heatmap distance loss for auto-segmentation. Mean Dice coefficients on the test dataset for dorsal, lateral, and ventral column and gray matter were 0.69, 0.67, 0.57, 0.54 on the left side and 0.68, 0.67, 0.59, 0.55 on the right side. The ROI-based mean FA value based on segmentation model strongly correlated with the value based on manual drawing. The percentages of the mean absolute error between the two values of multiple ROIs were 0.07, 0.07, 0.11, and 0.08 on the left side and 0.07, 0.1, 0.1, 0.11, and 0.07 on the right side. The proposed segmentation model has the potential to offer a more detailed spinal cord segmentation and would be beneficial for quantifying a more detailed status of the cervical spinal cord.

A Prediction Model for Normal Variation of Somatosensory Evoked Potential During Scoliosis Surgery.

Somatosensory evoked potential (SEP) has been commonly used as intraoperative monitoring to detect the presence of neurological deficits during scoliosis surgery. However, SEP usually presents an enormous variation in response to patient-specific factors such as physiological parameters leading to the false warning. This study proposes a prediction model to quantify SEP amplitude variation due to noninjury-related physiological changes of the patient undergoing scoliosis surgery. Based on a hybrid network of attention-based long-short-term memory (LSTM) and convolutional neural networks (CNNs), we develop a deep learning-based framework for predicting the SEP value in response to variation of physiological variables. The training and selection of model parameters were based on a 5-fold cross-validation scheme using mean square error (MSE) as evaluation metrics. The proposed model obtained MSE of 0.027[Formula: see text][Formula: see text] on left cortical SEP, MSE of 0.024[Formula: see text][Formula: see text] on left subcortical SEP, MSE of 0.031[Formula: see text][Formula: see text] on right cortical SEP, and MSE of 0.025[Formula: see text][Formula: see text] on right subcortical SEP based on the test set. The proposed model could quantify the affection from physiological parameters to the SEP amplitude in response to normal variation of physiology during scoliosis surgery. The prediction of SEP amplitude provides a potential varying reference for intraoperative SEP monitoring.

Machine Learning Prediction Models for Postoperative Stroke in Elderly Patients: Analyses of the MIMIC Database.

Objective: With the aging of populations and the high prevalence of stroke, postoperative stroke has become a growing concern. This study aimed to establish a prediction model and assess the risk factors for stroke in elderly patients during the postoperative period. Methods: ML (Machine learning) prediction models were applied to elderly patients from the MIMIC (Medical Information Mart for Intensive Care)-III and MIMIC-VI databases. The SMOTENC (synthetic minority oversampling technique for nominal and continuous data) balancing technique and iterative SVD (Singular Value Decomposition) data imputation method were used to address the problem of category imbalance and missing values, respectively. We analyzed the possible predictive factors of stroke in elderly patients using seven modeling approaches to train the model. The diagnostic value of the model derived from machine learning was evaluated by the ROC curve (receiver operating characteristic curve). Results: We analyzed 7,128 and 661 patients from MIMIC-VI and MIMIC-III, respectively. The XGB (extreme gradient boosting) model got the highest AUC (area under the curve) of 0.78 (0.75-0.81), making it better than the other six models, Besides, we found that XGB model with databalancing was better than that without data balancing. Based on this prediction model, we found hypertension, cancer, congestive heart failure, chronic pulmonary disease and peripheral vascular disease were the top five predictors. Furthermore, we demonstrated that hypertension predicted postoperative stroke is much more valuable. Conclusion: Stroke in elderly patients during the postoperative period can be reliably predicted. We proved XGB model is a reliable predictive model, and the history of hypertension should be weighted more heavily than the results of laboratory tests to prevent postoperative stroke in elderly patients regardless of gender.

Neural correlates of dynamic changes in working memory performance during one night of sleep deprivation.

Total sleep deprivation (TSD) is common in modern society leading to deterioration of multiple aspects of cognition. Dynamic interaction effect of circadian rhythmicity and homeostatic sleep pressure on sustained attention have been intensively investigated, while how this effect was represented on performance and cerebral responses to working memory, another important element of many neurobehavioral tasks, was not well elucidated. Thirty-six healthy subjects with intermediate chronotype performed the Sternberg working-memory task (SWMT) while undergoing functional magnetic resonance imaging every 2 hr from 10:00 p.m. on the first day to 6:00 a.m. on the second day. Using data from three imaging sessions (10:00 p.m., 04:00 a.m., and 06:00 a.m.), we found that the slowest SWMT reaction time and weakest cerebral responses were not at the end of TSD (06:00 a.m.) but during the early morning (04:00 a.m.) hours of the TSD. In addition, during this worst period of TSD, reaction time for the SWMT were found to be negatively correlated with task-related activation in the angular gyrus and positively correlated with the degree of negative correlation between the control and default networks. Our results revealed a rebound of SWMT reaction time and cerebral responses after the mid-time point of regular biological sleep night and provided more evidence that different cognitive tasks are differentially affected by sleep loss and circadian rhythmicity.

Decreased cortical and subcortical response to inhibition control after sleep deprivation.

The effects of sleep deprivation (SD) on the neural substrates of inhibition control are poorly understood. Here we used functional magnetic resonance imaging to examine the effects of 24 h of SD on cerebral activation during a stop-signal task in 20 normal young subjects. Behaviorally, subjects showed significantly delayed stop-signal reaction time (SSRT) following SD. In addition, reduced cerebral activation was found in the "stopping network" (including the inferior frontal gyrus [IFG], supplementary motor area, subthalamic nucleus [STN] and insula) and vision-related regions (occipital cortex, lingual gyrus and fusiform gyrus) after SD. These findings support the hypothesis that task-related activation in prefrontal cortex is particularly vulnerable to SD. After rested wakefulness (RW), significant negative correlations were found between SSRT and cerebral activation in left IFG, right hippocampus, right lingual gyrus, left STN and bilateral fusiform gyrus, with activation in left IFG making the most contribution. After SD, significant negative correlations were found between SSRT and activation in right middle frontal cortex, right IFG and left lingual gyrus, with the activation in right IFG making the most contribution. Furthermore, we observed significant interaction effects of state (SD or RW) with activation in bilateral IFG, left STN and left lingual gyrus on SSRT. In conclusion, sleep deprivation is associated with the deficits in inhibition-related neural activation and the altered correlation between SSRT and cerebral activation, especially in the bilateral IFG, left STN and left lingual gyrus.

"Practice makes perfect?" white matter microstructural characteristic predicts the degree of improvement in within-trial conflict processing across two weeks.

Several studies have investigated the trait-like characteristics of conflict processing at different levels. Our study extends these findings by reporting a practice-based improvement in within-trial conflict processing across two sessions. Eighty-three participants performed the same flanker task on two occasions 2 weeks apart. A subset of 37 subjects also underwent diffusion tensor imaging (DTI) scanning the day before the first behavioral task. Despite the trait-like characteristics of conflict processing, within-trial conflict processing in the second behavioral session was significantly shorter than that in the first session, indicating a practice-based improvement in conflict processing. Furthermore, changes in within-trial conflict processing across the two sessions exhibited significant individual differences. Tract-based spatial statistics revealed that the improvement across two sessions was related to the axial diffusivity values in white matter regions, including the body and splenium of the corpus callosum, right superior and posterior corona radiate, and right superior longitudinal fasciculus. Subsequently, lasso regression with leave-one-out cross validation was used to assess the predictive ability of white matter microstructural characteristics in significant regions. The results showed that 61% of individual variability in the improvement in the within-trial conflict processing could be explained by variations in the axial diffusivity values in the four significant regions and the within-trial conflict processing in the first session. These results suggest that axonal morphology in the white tracts connecting conflict-related regions predicts the degree of within-trial conflict processing improvement across two sessions.

Disrupted Resting-State Functional Connectivity in Hippocampal Subregions After Sleep Deprivation.

Previous studies have revealed that sleep deprivation (SD) alters hippocampal functional connectivity (FC). However, the effects of SD on the FC of hippocampal subregions are still unknown. In this study, we used a masked independent component analysis (mICA) to partition the hippocampus into several small regions and investigated the changes in the FC of each small region within the whole brain after 24 h of SD in 40 normal young subjects. First, we determined the optimal number of hippocampal subregions in a data-driven manner using a reproducibility analysis and chose 17 as the optimal number of hippocampal subregions. Second, we compared the FC of each subregion between rested wakefulness and SD states using mCIA. Reduced FC was found between the left junction of anterior and anterolateral hippocampal region and the default mode network and bilateral thalamus after SD (p < 0.05/17, Threshold-Free Cluster Enhancement correction, Bonferroni's corrected for the number of subregions). The FC between the left posterior of the anterolateral and the left lateral posterior of the anterior hippocampal regions and somatomotor network changed more negative after SD. However, increased FC was identified between the left middle hippocampal region and vision-related regions after SD. Our results reflect differential effects of SD on the FC in specific hippocampal regions and provide new insights into the impact of SD on the resting-state functional organization in the human brain.

Abnormal White Matter Microstructure in Lifelong Premature Ejaculation Patients Identified by Tract-Based Spatial Statistical Analysis.

INTRODUCTION: Several recent neuroimaging studies have identified functional and structural abnormalities in the cerebral cortex of lifelong premature ejaculation (LPE) patients, including task-related and resting-state brain function, and cortical thickness, although changes in white matter microstructure have not been reported. AIM: To assess the differences in white matter microstructure between LPE patients and healthy controls. METHODS: Diffusion tensor imaging (DTI) and tract-based spatial statistical analysis were used to detect differences in white matter microstructure between 32 LPE patients and 32 matched healthy controls. We also analyzed correlations of clinical indices with significant DTI-based features. MAIN OUTCOME MEASURES: DTI-based features (including fractional anisotropy [FA], mean diffusivity, axial diffusivity, and radial diffusivity) were assessed in LPE patients and controls, as well as the correlation of white matter changes in LPE patients with clinical data (including the premature ejaculation diagnostic tool score and the International Index of Erectile Function). RESULTS: LPE patients showed widespread increases in FA and axial diffusivity values compared with controls, including in the right posterior thalamic radiation, posterior corona radiata, bilateral posterior limb of the internal capsule, superior corona radiata, and external capsule. Further, FA in the right posterior thalamic radiation was positively correlated with the premature ejaculation diagnostic tool score in LPE patients. CLINICAL IMPLICATIONS: Changes of white matter microstructure may be an underlying marker for evaluating sensory conduction efficiency in LPE patients. STRENGTHS & LIMITATIONS: There are no previous studies examining white matter microstructure in LPE patients. The present study furthers our understanding of the etiology of LPE. Limitations include a cross-sectional study design without causal information, and no measurement of conduction efficiencies such as cortical somatosensory-evoked potential from the penis, or psychosocial factors. CONCLUSION: Our findings show potential microstructural white matter abnormalities related to LPE, suggesting that changes in fiber pathways connecting the cerebral cortex and the thalamus may play roles in the etiology of LPE. Gao M, Yang X, Liu L, et al. Abnormal White Matter Microstructure in Lifelong Premature Ejaculation Patients Identified by Tract-Based Spatial Statistical Analysis. J Sex Med 2018;15:1272-1279.

Prediction of the Effect of Sleep Deprivation on Response Inhibition via Machine Learning on Structural Magnetic Resonance Imaging Data.

Sleep deprivation (SD) impairs the ability of response inhibition. However, few studies have explored the quantitative prediction of performance impairment using Magnetic Resonance Imaging (MRI) data. In this study, structural MRI data were used to predict the change in response inhibition performance (ΔSSRT) measured by a stop-signal task (SST) after 24 h of SD in 52 normal young subjects. For each subject, T1-weighted MRI data were acquired and the gray matter (GM) volumes were calculated using voxel-based morphometry (VBM) analysis. First, the regions in which GM volumes correlated with ΔSSRT were explored. Then, features were extracted from these regions and the prediction process was performed using a linear regression model with four-fold cross-validation. We found that the GM volumes of the left middle frontal gyrus (L_MFG), pars opercularis of right inferior frontal gyrus (R_IFG), pars triangularis of left inferior frontal gyrus, pars opercularis of right rolandic area, left supplementary motor area (L_SMA), left hippocampus, right lingual gyrus, right postcentral gyrus and left middle temporal gyrus (L_MTG) could predict the ΔSSRT with a low mean square error of 0.0039 ± 0.0011 and a high Pearson's correlation coefficient between the predicted and actual values of 0.948 ± 0.0503. In conclusion, our results demonstrated that a linear combination of structural MRI data could accurately predict the change in response inhibition performance after SD. Further studies with larger sample sizes and more comprehensive sample may be necessary to validate these findings.

Dynamics of cerebral responses to sustained attention performance during one night of sleep deprivation.

Total sleep deprivation (TSD) is increasingly common in modern society bringing various neurobehavioural effects. Dynamic changes of behaviour performances during TSD have been reported extensively, while the cerebral activation underlying such changes have not been elucidated clearly. This study aimed to investigate dynamic changes in cerebral responses to the fastest and slowest psychomotor vigilance task (PVT) trials during TSD. Thirty-six healthy subjects with intermediate chronotype performed the PVT while undergoing functional magnetic resonance imaging every 2 h from 22:00 hours on the first day to 06:00 hours on the second day (i.e. 22:00, 12:00, 02:00, 04:00 and 06:00 hours; a total of five imaging sessions). Behaviourally, significant time effects were found for the PVT performance. For imaging results, significant activation alterations were found in the cognitive control network and the default mode network (DMN) for the fastest and slowest PVT trials, respectively. Time-course analysis indicated that the largest differences for behavioural results and imaging results happened in session 4 and became more prominent in session 5. Our findings provide more detailed information about the process of sustained attention activation during one night of TSD and add information regarding the effect of circadian rhythmicity and homeostatic sleep pressure on regional brain responses.

Alterations in cortical thickness in nonmedicated premature ejaculation patients: A morphometric MRI study.

PURPOSE: Premature ejaculation (PE) is one of the most common sexual dysfunctions in men. However, there has been little research evaluating alterations in brain structure related to PE. We aimed to investigate the characteristics of nonmedicated PE patients in terms of brain morphometry. MATERIALS AND METHODS: The sample consisted with 32 medication-naïve adult men with clinical diagnosed PE and matched 31 healthy controls. All participants received diagnostic interviews and 3.0 Tesla MRI scans. Automatic segmentation processing of MRI structure images was performed using FreeSurfer software and cerebral cortical thickness between groups was compared. RESULTS: The PE group had thicker cortex in widespread regions, including the frontal, parietal and occipital lobe, and limbic system, compared with the healthy control group (P < 0.05). Moreover, the duration is negatively correlated with the mean cortical thickness of the right medial orbitofrontal cortex, right precentral gyrus and left superior frontal cortex (R2 = 0.29, P < 0.003; R2 = 0.163, P < 0.04; R2 = 0.2, P < 0.02), while the Premature Ejaculation Diagnostic Tool score is negatively correlated with the mean cortical thickness of the left caudal middle frontal cortex (R2 = 0.33, P < 0.005). CONCLUSION: The result highlights the structural features of PE and suggests the relationship with the severity of impairment is related to the severity of anatomic abnormality with the relevant brain region. These results support the value of imaging measures as markers for understanding the physiopathology of PE. LEVEL OF EVIDENCE: 1 Technical Efficacy: Stage 3 J. Magn. Reson. Imaging 2018;47:656-662.

White Matter Microstructural Properties are Related to Inter-Individual Differences in Cognitive Instability after Sleep Deprivation.

Several diseases are characterized by cognitive instability, which is amplified in the conditions of sleep deprivation (SD). Cognitive instability in SD can be examined by the number of lapses on the psychomotor vigilance test (PVT), which is considered to be a gold standard in the field. However, the number of PVT lapses widely range according to inter-individual differences, from apparent cognitive resistance to severe cognitive impairment. In this study, tract-based spatial statistical analyses with multiple diffusion tensor imaging-derived characteristics (i.e., fractional anisotropy (FA), mean diffusivity, radial diffusivity, and axial diffusivity) were employed to investigate the relationships between the number of PVT lapses and the diffusion characteristics. A hierarchical linear regression model was then used to assess the contributions of tract-specific FA values in predicting PVT lapses. Finally, dichotomized analysis was used to investigate white matter (WM) differences between resilient and vulnerable groups. Our results showed significant negative correlations between numbers of PVT lapses and FA in multiple WM tracts, with the FA variations in the superior longitudinal fasciculus and splenium of the corpus callosum accounting for nearly 37.5% of individual variability in PVT lapses. In addition, dichotomized analyses indicated that the resilient participants exhibited significantly higher FA values compared with the vulnerable participants. Together, these findings suggest that cognitive instability after SD was closely associated with individual differences in WM integrity.

Brain white matter fiber tracts involved in post-transjugular intrahepatic portosystemic shunt hepatic myelopathy.

We aimed to detect alterations in diffusion characteristics of brain white matter in hepatic myelopathy (HM) patients. Liver cirrhosis patients with (n=25) and without (n=18) HM after transjugular intrahepatic portosystemic shunt and 26 healthy controls were enrolled in this study. All participants were scanned with diffusion tensor imaging on a 3T Siemens scanner. Tract-based spatial statistics analysis was used to detect abnormalities of intracranial white matter tracts. Correlations between clinical characteristics and diffusion metrics were also calculated. HM patients showed widespread decreased fractional anisotropy values in association fibers, callosal fibers, thalamic fibers, and limbic system fibers (P<0.01, family-wise error-corrected) compared with healthy controls. In addition, HM patients showed lower fractional anisotropy values in the corpus callosum, corona radiata, external capsule, and superior longitudinal fasciculus compared with cirrhosis patients without myelopathy (P<0.01, family-wise error-corrected). Furthermore, limb muscle strength grading was correlated with the diffusion characteristics of the corpus callosum and superior longitudinal fasciculus in HM patients (P<0.05). HM patients suffer from more distinct changes of white matter fiber tracts than cirrhosis patients without myelopathy. In addition, alterations of the corpus callosum and superior longitudinal fasciculus may be associated with the major motor disturbance in HM. Our finding may shed light on the underlying neuropathological mechanism of HM.

Effects of the Brain-Derived Neurotrophic Factor Val66Met polymorphism and resting brain functional connectivity on individual differences in tactile cognitive performance in healthy young adults.

Cognitive processes involve input from multiple sensory modalities and obvious differences in the level of cognitive function can be observed between individuals. Evidence to date understanding the biological basis of tactile cognitive variability, however, is limited compared with other forms of sensory cognition. Data from auditory and visual cognition research suggest that variations in both genetics and intrinsic brain function might contribute to individual differences in tactile cognitive performance. In the present study, by using the tactual performance test (TPT), a widely used neuropsychological assessment tool, we investigated the effects of the brain-derived neurotrophic factor (BDNF) Val66Met polymorphism and resting-state brain functional connectivity (FC) on interindividual variability in TPT performance in healthy, young Chinese adults. Our results showed that the BDNF genotypes and resting-state FC had significant effects on the variability in TPT performance, together accounting for 32.5% and 19.1% of the variance on TPT total score and Memory subitem score respectively. Having fewer Met alleles, stronger anticorrelations between left posterior superior temporal gyrus and somatosensory areas (right postcentral gyrus and right parietal operculum cortex), and greater positive correlation between left parietal operculum cortex and left central opercular cortex, all correspond with better performance of TPT task. And FC between left parietal operculum cortex and left central opercular cortex might be a mediator of the relationship between BDNF genotypes and Memory subitem score. These data demonstrate a novel contribution of intrinsic brain function to tactile cognitive capacity, and further confirm the genetic basis of tactile cognition. Our findings might also explain the interindividual differences in cognitive ability observed in those who are blind and/or deaf from a new perspective.