Myocardial Infarction

Background

The most common and cost-effective approach to diagnose myocardial ischemia/infarction is to use electrocardiography (ECG). Another, more expensive, diagnostic techniques are biomarkes (e.g. troponin T or I, or CK-MB) and medical imagaging (e.g echocardiography). In addition, biomarkers could delayed a treatment when patient is at risk, i.e. in the acute stage of myocardial ischemia. The former approach, ECG, is an electrical representation of the hearth and is acquired using special device called electrocardiograph. For the standard ECG a 12-lead system is used. Each lead represents a projection of the hearth vector to a particular direction. The myocardial ischemia and infarction cause morphological changes at the ECG.

ECG limb leads ECG precordial leads
Figure 1a. Limb leads with inferior myocardial infarction. Figure 1b . Precordial leads with inferior myocardial infarction.

Decision rules are used to assess morphological changes at ECG caused by myocardial ischemia and infarction (ST-T changes for acute state, Q and T wave changes for an infarcted myocardium). Rules were suggested by cardiologists and originated from theoretical foundations and experience. It were adopted later into computerized scoring/coding/decision systems (hereinafter commonly referred as decision systems); the best known are: the Selvester score, the Novacode, and Siemens 440/740.

Aim of the project

The aim of this project is to automatically detect the morphological changes of ECG caused by myocardial ischemia/infarction and diagnose where and when these changes occured, i.e. time and location. The automatic classification will serve as decission support for doctors and help them with diagnosis and assesment of ECG.

The overall methodology
Figure 2. The overall scheme of methodology.

We utilize the best known scoring/coding/decision systems (the Selvester QRS score, the Novacode, and the Siemens 440/740) and several learning algorithms (Ripper, C4.5, and SVM). The decision systems were developed with different purposes (the Selvester for estimation of MI size, the Novacode for clinical and epidemiologic studies, and the Siemens for ECG device Siemens 440/740). In this work we combined these systems with additional simple rules, created by AdaBoost, and compared performance to: (i) decision systems alone, (ii) base classifiers (Ripper, C4.5, and SVM). The overall methodology is shown in Figure 2.

Contact persons: Jiri Spilka

Cooperating institutions
  • Medical Technologies CZ, Czech republic

Funding
  • This project is supported by the the CVUT Grant SGS10/279/OHK3/3T/13.


Previous page: Cardiotocography
Next page: Holter Data Classification