LETI: A Standard Developed at LETI Will Improve the Accuracy of Heart Diagnostics
Researchers from LETI and Almazov Center created a verified database of synchronous records of heart rhythm to improve the accuracy of modern diagnostic systems for cardiac abnormalities.
According to the World Health Organization, cardiac rhythm disorders (tachyarrhythmia, atrial fibrillation, etc.) affect every third cardiac patient. To develop means and methods of diagnostics of such disorders, electrocardiogram (ECG) databases are used, which contain structured and digitized information about patients with various heart rhythm disorders. These data are the benchmark that helps devices correctly determine whether or not a person has cardiac abnormalities.
Today, several databases of ECG traces are actively used to ensure the precision of diagnostic devices and algorithms. However, in the last decade, high-resolution ECG technologies, for which previously there was no database of verified data, have developed significantly.
“We have created a unique database of verified synchronous high-resolution ECG traces and, most importantly, intracardiac electrograms. The database contains data from a large number of patients with dangerous heart rhythm abnormalities. We verified these records using various methods, and, therefore, they have a great value for evaluation of modern diagnostic tools and prediction of heart rate disturbances in comparison with analogs”.
Zafar Yuldashev, Head of the Department of Bioengineering Systems at LETI
Researchers created the database using many years of anonymized records from the patients of the Almazov National Medical Research Centre. The intracardiac electrogram was recorded using special electrodes that were inserted inside the heart through the vena cava. These data were then synchronized with external ECG traces to achieve high accuracy. Episodes of normal heart rhythm and abnormalities were marked on the recordings, after which a cardiologist checked the correctness of the results.
The project resulted in a database consisting of two-second recordings of cardiac muscle abnormalities and longer fragments ranging from 30 seconds to several minutes. The latter include a long recording of a normal heartbeat with inclusions of short episodes of arrhythmia: it is required to be able to detect such fragments for all modern diagnostic systems.
“Our database is important primarily to researchers and engineers who develop new algorithms and equipment for diagnosing dangerous heart rhythm disorders. We have already received state registration for the database in Russia and are now planning to do the same in the USA,” notes Zafar Yuldashev.
The database is arranged as a table and is openly accessible to any researcher. It contains the patient’s age, sex, and diagnosis, and all data are structured by type of abnormality. The researchers also filtered out all extraneous noise and interference that occurred during the recording of cardiograms.