Суъний нейрон тармоклари асосида миокард инфарктни таснифлаш ва жойлашув урнини аниклаш
Аннотация
В этой статье была разработана архитектура CNN на основе электрокардиограммы (ЭКГ) для определения и классификации местоположения инфаркта миокарда. База данных была разработана на основе системы ЭКГ с 12 отведениями. База данных, состоящая из 10 классов инфаркта миокарда и исходных данных от здорового человека, была обучена архитектуре CNN, и результат достиг 98,47%.
Ушбу маколада электрокардиограммага (ЭКГ) асосан миокард инфарктни жойлашув урнини аникловчи ва таснифловчи CNN(convolution neural network, CNN) архитектураси ишлаб чикилди. 12-каналли ЭКГ тизимига асосан маълумотлар базаси лойих,алаштирилди. CNN архитектурасида миокард инфарктнинг 10 синфи ва согаом одамнинг дастлабки маълумотларидан ташкил топган база укитилди ва олинган натижа 98.47% га етди.
In this paper, has been developed CNN architecture based on electrocardiogram (ECG)to determine and classify the location of myocardial infarction. Some approaches for detecting and classifying of myocardial infarction based on CNN networks by global researchers have discussed. The methods and means of detection of myocardial infarction have been studied. Deep learning techniques for classifying myocardial infarction were analyzed, including CNN. Current study focused primarily on the classification of myocardial infarction based on ECG images. The database was designed based on the 12-lead ECG system. A database consisting of 10 classes of myocardial infarction and baseline data from a healthy person was trained in CNN architecture. Scientists’ work on automatic detection and classification of myocardial infarction has been studied and the results compared. Eleven classes created based on *.mat files for MI classification and each *.mat file consisted of amount of different data. For each class, network accuracy was calculated separately and the average of the results determined. The results were compared based on the table with the results of other researchers taken from their works. The specifics of each of taken results were also discussed.The results of the article are summarized and future work is planned. The result obtained by the proposed method was 98.47%.
Ушбу маколада электрокардиограммага (ЭКГ) асосан миокард инфарктни жойлашув урнини аникловчи ва таснифловчи CNN(convolution neural network, CNN) архитектураси ишлаб чикилди. 12-каналли ЭКГ тизимига асосан маълумотлар базаси лойих,алаштирилди. CNN архитектурасида миокард инфарктнинг 10 синфи ва согаом одамнинг дастлабки маълумотларидан ташкил топган база укитилди ва олинган натижа 98.47% га етди.
In this paper, has been developed CNN architecture based on electrocardiogram (ECG)to determine and classify the location of myocardial infarction. Some approaches for detecting and classifying of myocardial infarction based on CNN networks by global researchers have discussed. The methods and means of detection of myocardial infarction have been studied. Deep learning techniques for classifying myocardial infarction were analyzed, including CNN. Current study focused primarily on the classification of myocardial infarction based on ECG images. The database was designed based on the 12-lead ECG system. A database consisting of 10 classes of myocardial infarction and baseline data from a healthy person was trained in CNN architecture. Scientists’ work on automatic detection and classification of myocardial infarction has been studied and the results compared. Eleven classes created based on *.mat files for MI classification and each *.mat file consisted of amount of different data. For each class, network accuracy was calculated separately and the average of the results determined. The results were compared based on the table with the results of other researchers taken from their works. The specifics of each of taken results were also discussed.The results of the article are summarized and future work is planned. The result obtained by the proposed method was 98.47%.