Morning blood pressure surge and intracranial atherosclerotic plaque characteristics: a high-resolution magnetic resonance vessel wall imaging study.

OBJECTIVE: To investigate the relationship between morning blood pressure surge (MBPS) and intracranial atherosclerotic plaque burden and vulnerability. METHODS: A total of 267 ischaemic stroke patients were retrospectively analysed. Sleep-trough and prewaking MBPS were calculated from ambulatory blood pressure monitoring (ABPM). Plaque characteristics, including intraplaque haemorrhage (IPH), maximum wall thickness (max WT), and stenosis degree, were obtained from high-resolution MR vessel wall imaging (HR-vwMRI). Linear and logistic regression were used to detect the association. RESULTS: Subjects with the top tertile of sleep-trough MBPS (≥15.1 mmHg) had a lower prevalence (9.1% vs. 19.6%, P = .029) of severe stenosis (≥70%) than others. Subjects within the top tertile of prewaking MBPS (≥7.6 mmHg) had a lower percentage of IPH (27.3% vs. 40.4%, P = .035) than others. After adjusting for stroke risk factors (age, sex, diabetes, hyperlipidaemia, hyperhomocysteinaemia, smoking, and family stroke history) and 24-h mean systolic blood pressure, 10 mmHg sleep-trough MBPS increment was associated with 0.07mm max WT reduction, and the top tertile MBPS group was associated with a lower chance of severe stenosis (odd ratio = 0.407, 95% CI, 0.175-0.950). Additionally, an increased prewaking MBPS is associated with a lower incidence of IPH, with OR = 0.531 (95% CI, 0.296-0.952). Subgroup analysis demonstrated that the positive findings could only be seen in non-diabetic subjects. CONCLUSION: Increment of MBPS is negatively associated with intracranial atherosclerotic plaque burden and vulnerability, and this relationship remains significant in the non-diabetic subgroup. ADVANCES IN KNOWLEDGE: This study provided evidence that MBPS was associated with the intracranial atherosclerotic plaque burden and vulnerability on HR-vwMRI.

Bidirectional Recurrent Neural Network And Convolutional Neural Network (BiRCNN) For ECG Beat Classification.

We propose a novel electrocardiogram (ECG) beat classification algorithm using a combination of Bidirectional Recurrent Neural Network (BiRNN) and Convolutional Neural Network (CNN) named as BiRCNN. Our model is an end-to-end model. The morphological features of each ECG beat is extracted by CNN. Then the features of each beat are considered in the context via BiRNN. The assessment on MIT-BIH Arrhythmia Database (MITDB) resulted in a sensitivity of 98.7% and a positive predictivity of 96.4% on average for the VEB class. For the SVEB class, the sensitivity was 92.8%, which was an over 6% promotion compared with the state-of-the-art method, and the positive predictivity was 81.9% on average. The results demonstrate the superior classification performance of our method.

Region Aggregation Network: Improving Convolutional Neural Network for ECG Characteristic Detection.

Detection of ECG characteristic points serves as the first step in automated ECG analysis techniques. We propose a novel end-to-end deep learning scheme called Region Aggregation Network (RAN) for ECG characteristic points de- tection. A 1D Convolutional Neural Network (CNN) is adopted to automatically process ECG signals. A novel strategy of Region Aggregation is proposed to replace the conventional fully connected layer as regressor. Our work provides robust and accurate detection performance on public ECG database. The evaluation results of our method on QT database show comparable detection accuracy compared with state-of-the-art works.