Journal: International Journal of Cardiology. Cardiovascular Risk and Prevention

Article Title: Prevalence, associated factors, and prognostic value of P wave abnormality in patients with coronary artery disease

doi: 10.1016/j.ijcrp.2025.200533

Figure Lengend Snippet: (A) Representative electrocardiogram for determining PTFV1. PTFV1 (mm∙s) was calculated by the duration (s) of the terminal negative part of the P wave in lead V1 multiplied by the absolute value of its amplitude (mm). (B) Distribution of PTFV1 in patients with CAD who underwent PCI. The red bars indicate abnormal PTFV1, and the blue bars show normal PTFV1. CAD = coronary artery disease, PCI = percutaneous coronary intervention, and PTFV1 = P-wave terminal force in V1. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Article Snippet: These parameters were also automatically analyzed by the MBF-1000 ECG Data Management System (Fukuda Denshi Co., Ltd., Tokyo, Japan) in 1179 patients.

Techniques:

Journal: JACC Asia

Article Title: Deep Learning-Based Identification of Echocardiographic Abnormalities From Electrocardiograms

doi: 10.1016/j.jacasi.2024.10.012

Figure Lengend Snippet: Flow Chart Showing the Study Datasets The paired electrocardiogram (ECG) and echocardiography data collected from 8 centers were used in 6 data sets, namely, Mitsui Memorial Hospital (Mitsui), Asahi General Hospital (Asahi), Sakakibara Heart Institute (Sakakibara), Jichi Medical University Saitama Medical Center (Jichi), Tokyo Bay Urayasu Ichikawa Medical Center (TokyoBay), and JR Tokyo General Hospital (JR), for model development, and in 2 data sets, namely, The University of Tokyo Hospital (UTokyo) and NTT Medical Center Tokyo (NTT), for external validation. The data sets for model development were split further into a training set, a validation set, and a test set.

Article Snippet: ECG data from Fukuda Denshi (Tokyo, Japan) were used for UTokyo and Mitsui while ECG data from Nihon Kohden (Tokyo, Japan) were used for Asahi, Sakakibara, Jichi, TokyoBay, JR, and NTT.

Techniques:

Journal: JACC Asia

Article Title: Deep Learning-Based Identification of Echocardiographic Abnormalities From Electrocardiograms

doi: 10.1016/j.jacasi.2024.10.012

Figure Lengend Snippet: Overview of the Study Twelve echocardiographic finding labels for left-sided cardiac abnormalities, valvular heart diseases, and right-sided cardiac abnormalities were assigned from paired ECGs and echocardiograms. These labeled data sets were trained using convolutional neural network (CNN) to generate models for each specific echocardiographic finding. Subsequently, logistic regression was used on the output from these CNN models to predict the composite findings label. AR = aortic regurgitation; AS = aortic stenosis; DD = diastolic dysfunction; ECG = electrocardiogram; echo = echocardiographic; LAD = left atrial dilatation; LVD = left ventricular dilatation; LVEF = left ventricular ejection fraction; LVH = left ventricular hypertrophy; MR = mitral regurgitation; PH = pulmonary hypertension; RVD = right ventricular dysfunction; TR = tricuspid regurgitation; WMA = wall motion abnormality.

Article Snippet: ECG data from Fukuda Denshi (Tokyo, Japan) were used for UTokyo and Mitsui while ECG data from Nihon Kohden (Tokyo, Japan) were used for Asahi, Sakakibara, Jichi, TokyoBay, JR, and NTT.

Techniques: Labeling

Journal: JACC Asia

Article Title: Deep Learning-Based Identification of Echocardiographic Abnormalities From Electrocardiograms

doi: 10.1016/j.jacasi.2024.10.012

Figure Lengend Snippet: Patient Demographic and Clinical Characteristics

Article Snippet: ECG data from Fukuda Denshi (Tokyo, Japan) were used for UTokyo and Mitsui while ECG data from Nihon Kohden (Tokyo, Japan) were used for Asahi, Sakakibara, Jichi, TokyoBay, JR, and NTT.

Techniques: