A5078373751- Cardiac pacing and defibrillation studies
- ECG Monitoring and Analysis
- Cardiac electrophysiology and arrhythmias
- Cardiac Arrhythmias and Treatments
- Hemodynamic Monitoring and Therapy
- Non-Invasive Vital Sign Monitoring
- EEG and Brain-Computer Interfaces
- Cardiovascular Health and Disease Prevention
- Phonocardiography and Auscultation Techniques
- Neurological disorders and treatments
- Electrical and Bioimpedance Tomography
- Sensor Technology and Measurement Systems
- Heart Rate Variability and Autonomic Control
- Cardiovascular Function and Risk Factors
- Atrial Fibrillation Management and Outcomes
- Analog and Mixed-Signal Circuit Design
- Blood Pressure and Hypertension Studies
- Brain Tumor Detection and Classification
- Ion channel regulation and function
- COVID-19 diagnosis using AI
- Neural dynamics and brain function
- Heart Failure Treatment and Management
- Neuroscience and Neural Engineering
- Functional Brain Connectivity Studies
- Machine Learning in Healthcare
Czech Academy of Sciences, Institute of Scientific Instruments
2016-2025
Czech Academy of Sciences
2011
Nonselective His-bundle pacing (nsHBp), nonselective left bundle branch (nsLBBp), and ventricular septal myocardial (LVSP) are recognized as physiological techniques.The purpose of this study was to compare differences in depolarization between these techniques using ultra-high-frequency electrocardiography (UHF-ECG).In patients with bradycardia, nsHBp, nsLBBp (confirmed concomitant [LBB] capture), LVSP (pacing [LV] position without proven LBB capture) were performed. Timings activations...
Abstract Manual and semi-automatic identification of artifacts unwanted physiological signals in large intracerebral electroencephalographic (iEEG) recordings is time consuming inaccurate. To date, unsupervised methods to accurately detect iEEG are not available. This study introduces a novel machine-learning approach for detection clinically controlled conditions using convolutional neural networks (CNN) benchmarks the method’s performance against expert annotations. The method was trained...
Left bundle branch pacing (LBBP) produces delayed, unphysiological activation of the right ventricle. Using ultra-high-frequency electrocardiography (UHF-ECG), we explored how bipolar anodal septal with direct LBB capture (aLBBP) affects resultant ventricular depolarization pattern.In patients bradycardia, His (HBP), unipolar nonselective LBBP (nsLBBP), aLBBP, and (RVSP) were performed. Timing local activation, in leads V1-V8, was displayed using UHF-ECG, electrical dyssynchrony (e-DYS)...
This paper introduces a winning solution (team ISIBrno-AIMT) to the PhysioNet Challenge 2021. The method is based on ResNet deep neural network architecture with multi-head attention mechanism for ECG classification into 26 independent groups. model optimized using mixture of loss functions, i.e., binary cross-entropy, custom challenge score function, and sparsity function. Probability thresholds each class are estimated evolutionary optimization method. final consists three submodels...
Electrical cardioversion presents one of the treatment options for atrial fibrillation (AF). However, early recurrence rate is high, reaching ~40% three months after procedure. Features based on vectorcardiographic signals were explored to find association with AF. Eighty-four patients non-paroxysmal AF referred electrical prospectively studied; was present in 40 (47.6%). Patients underwent 24-h Holter ECG monitoring procedure assess recurrence. Pre-procedural 12-lead ECGs (10 s, 1 kHz)...
The automated detection of arrhythmia in a Holter ECG signal is challenging task due to its complex clinical content and data quantity. It also the fact that usually affected by noise. Such noise may be result regular activity patients using ECG—partially unplugged electrodes, short-time disconnections movement, or disturbances caused electric devices infrastructure. Furthermore, patient activities such as movement affect signals and, connection with artificial noise, render non-readable...
This paper describes a method for automated discrimination of heart sounds recordings according to the Physionet Challenge 2016. The goal was decide if recording refers normal or abnormal it is not possible (i.e. 'unsure' recordings).Heart S1 and S2 are detected using amplitude envelopes in band 15-90 Hz. averaged shape S1/S2 pair computed from five different bands (15-90 Hz; 55-150 100-250 200-450 400-800 Hz). A total 53 features extracted data. largest group statistical properties shapes;...
The present study introduces a new ultra-high-frequency 14-lead electrocardiogram technique (UHF-ECG) for mapping ventricular depolarization patterns and calculation of novel dyssynchrony parameters that may improve the selection patients application cardiac resynchronization therapy (CRT).Components ECG in sixteen frequency bands within 150 to 1000 Hz range were used create maps. maximum time difference between UHF QRS complex centers mass leads V1 V8 was defined as electrical (e-DYS),...
Background: Three different ventricular capture types are observed during left bundle branch pacing (LBBp). They selective LBB (sLBBp), non-selective (nsLBBp), and myocardial septal transiting from nsLBBp while decreasing the output (LVSP). Study aimed to compare differences in depolarization between these captures using ultra-high-frequency electrocardiography (UHF-ECG). Methods: Using decremental voltage output, we identified studied nsLBBp, sLBBp, LVSP patients with bradycardia. Timing of...
Abstract Background Right ventricular (RV) pacing causes delayed activation of remote segments. We used the ultra‐high‐frequency ECG (UHF‐ECG) to describe depolarization when different RV locations. Methods In 51 patients, temporary was performed at septum (mSp); further subclassified as right inflow tract (RVIT) and outflow (RVOT) for septal positions (below or above plane His bundle in anterior oblique), apex, lateral wall, basal with nonselective RBB capture (nsHBorRBBp). The timings...
Abstract While various QRS detection and classification methods were developed in the past, Holter ECG data acquired during daily activities by wearable devices represent new challenges such as increased noise artefacts due to patient movements. Here, we present a deep-learning model detect classify complexes single-lead ECG. We introduce novel approach, delivering one inference step. used private dataset (12,111 recordings, length of 30 s) for training, validation, testing method. Twelve...
The effect of left ventricular myocardial septal (LVSP) and bundle branch pacing (LBBP) on synchrony LV hemodynamics is poorly understood. To investigate the impact LVSP LBBP versus biventricular (BVP) electrical in CRT patients. In candidates with conduction disease, was assessed by measuring QRSd using ultra-high-frequency ECG (UHF-ECG). (lv-DYS) as a difference between first activation V1-V8 to last from V4-V8. estimated invasive systolic blood pressure measurement during multiple...
Background: Although cardiac resynchronization therapy (CRT) is beneficial in heart failure patients with left bundle branch block, 30% of these do not respond to the therapy. Identifying before implantation device one current challenges clinical cardiology. Methods: We verified diagnostic contribution and an optimized computerized approach measuring ventricular electrical activation delay (VED) from body surface 12-lead ECGs. applied method ECGs acquired (baseline) MADIT-CRT trial...
Abstract EEG signal processing is a fundamental method for neurophysiology research and clinical neurology practice. Historically the classification of into physiological, pathological, or artifacts has been performed by expert visual review recordings. However, size data recordings rapidly increasing with trend higher channel counts, greater sampling frequency, longer recording duration complete reliance on not sustainable. In this study, we publicly share annotated intracranial clips from...
Objective. This paper introduces a winning solution (team ISIBrno-AIMT) to the official round of PhysioNet Challenge 2021. The main goal challenge was classification ECG recordings into 26 multi-label pathological classes with variable number leads (e.g. 12, 6, 4, 3, 2). objective this study is verify whether multi-head-attention mechanism influences model performance.Approach. We introduced an method based on ResNet architecture multi-head attention for challenge. However, empirical...
The electroencephalogram (EEG) is a cornerstone of neurophysiological research and clinical neurology. Historically, the classification EEG as showing normal physiological or abnormal pathological activity has been performed by expert visual review. potential value unbiased, automated long recognized, in recent years application machine learning methods received significant attention. A variety solutions using convolutional neural networks (CNN) for have emerged with impressive results....
Atrial fibrillation (AF) is a disease affecting 1-2 % of the population.Due to its episodic behavior, it usually detected using Holter recordings.While various AF detection methods have been described in past, still remains problematic because holter recordings may contain other arrhythmias (OA) and, moreover, they be influenced by patient movements.In accordance with Physionet Challenge 2017, we propose an autonomous and robust method distinguishing between pathological normal...
Cardiac resynchronization therapy (CRT) is an effective treatment that reduces mortality and improves cardiac function in patients with left bundle branch block (LBBB). However, about 30% of passing the current criteria do not benefit or only a little from CRT. Three predictors based on different ECG properties were compared: 1) "strict" classification (SLBBB); 2) QRS area; 3) ventricular electrical delay (VED) which defines septal-lateral conduction delay. These have never been analyzed...
The patients with the long QT syndrome type-1 (LQT-1) have an impaired adaptation of interval to heart rate changes. Yet, description dynamic QT-RR coupling in genotyped LQT-1 has never been thoroughly investigated.We propose a method model by defining transfer function characterizing relationship between and its previous RR intervals measured from ambulatory Holter recordings. Three parameters are used characterize coupling: fast gain (Gain(F) ), slow (Gain(L) time constant (τ). We...
Abstract The design, properties, and possible diagnostic contribution of a multichannel bioimpedance monitor (MBM) with three independent current sources are presented in this paper. simultaneous measurement at 18 locations (the main part the body, legs, arms, neck) provides completely new information, on basis which more precise haemodynamic parameters can be obtained. application MBM during various stages, such as resting supine position, tilting, exercise stress, respiration manoeuvres,...
Cardiac diseases are the most common cause of death.The fully automated classification electrocardiogram (ECG) supports early capturing heart disorders, and, consequently, may help to get treatment early.Here in this paper, we introduce a deep neural network for human ECG into 24 independent groups, example, atrial fibrillation, 1st degree AV block, Bundle branch blocks, premature contractions, changes ST segment, normal sinus rhythm, and others.The architecture utilizes convolutional with...
Abstract The study introduces and validates a novel high-frequency (100–400 Hz bandwidth, 2 kHz sampling frequency) electrocardiographic imaging (HFECGI) technique that measures intramural ventricular electrical activation. Ex-vivo experiments clinical measurements were employed. Ex-vivo, two pig hearts suspended in human-torso shaped tank using surface electrodes, epicardial electrode sock, plunge electrodes. We compared conventional (ECGI) with activation by HFECGI verified sock Clinical...