Localization of Origins of Premature Ventricular Contraction by Means of Convolutional Neural Network from 12-lead ECG
We proposes a novel method to localize origins of premature ventricular contractions (PVCs) from 12-lead ECG using convolutional neural network (CNN) and a realistic computer heart model. The proposed method consists of two CNNs to classify among ventricular sources from 25 segments and from epicardium (Epi) or endocardium (Endo). …
Electrical Properties Tomography Based on B1 Maps in MRI: Principles, Applications, and Challenges
We review the basic principle of electrical properties tomography (EPT), reconstruction methods, biomedical applications including tumor imaging, and existing challenges. As an important application of EPT, the estimation of specific absorption rate (SAR) and its current development will also be discussed.
Noninvasive Electromagnetic Source Imaging and Granger Causality Analysis: An Electrophysiological Connectome (eConnectome) Approach
We present an electrophysiological connectome (eConnectome) approach to study underlying brain networks in a noninvasive manner. This approach was directly tested by estimating epileptic networks from EEG/MEG measurements in patients suffering from focal epilepsy. The results obtained from the proposed approach were consistent with invasive clinical findings, in these patients.
Electrophysiological Source Imaging of Brain Networks Perturbed by Low-Intensity Transcranial Focused Ultrasound
We report an experimental investigation to noninvasively detect electrophysiological response induced by low-intensity transcranial focused ultrasound (tFUS) in an in vivo animal model, and perform electrophysiological source imaging (ESI) of tFUS-induced brain activity from noninvasive scalp EEG recordings. Neural activation has been observed following low-intensity tFUS, for various ultrasound intensities and sonication durations.
EEG Source Imaging Enhances the Decoding of Complex Right Hand Motor Imagery Tasks
We extend previous EEG source imaging (ESI) work to decode four natural MI tasks of the right hand that may be utilized for realistic and intuitive BCI control. We report an increase of up to 18.6% for individual task classification, and an increase of 12.7% for the overall classification using the proposed ESI approach over the traditional sensor-based method.