Electrocardiographic analysis for telemedicine based on a low-cost software approach

Introduction

Trends in telemedicine applied for clinical cardiology involve techniques such as the transmission of Holter monitoring recordings, as well as of electrocardiographic (ECG) measurements in the ICU. Both techniques enable more efficient clinical procedures, especially in the case of acute myocardial infarction [1]. However, this transmission is a complex task, since different file formats must be handled by the system, thus requiring conversion from one format to the other.

Objective

To propose a low-cost software tool for the analysis of ECG recordings of different file formats, which enable processing of data measured in the ICU or arising from telemedicine systems.

Materials and methods

Twenty ECG recordings, taken from acute myocardial infarction patients by a telemedicine system, were stored as .pdf files. We have considered just the derivations Avf, II and III. In the following, they were analysed by our software tool to estimate the variation of the area of the ST segment. First, the .pdf files were converted to eight-bit .bmp files, as images with 2000 × 1600 pixel resolution, in a grey scale. Axes are defined in the image to extract a data file, which may be converted into any desired format. This data file was processed by two different methods, in order to estimate the variation of the ST-segment area. The first method employs vertical axis scanning whereas the second method carries out standard integration, based on a finite-interval technique.

Results

The two methods presented the following average errors for the estimation: 1.80% for derivation Avf, 1.27% for derivation II and 1.38% for derivation III. Average variances were 7000 for method 1 (axis scanning) and 7200 for method 2 (standard integration).

Conclusion

Simple and cost-effective methods for the analysis of ECG recordings arising from a telemedicine system, which may be used in current computer configurations in the ICU, were successfully employed to estimate the variation of the ST-segment area for myocardial infarction patients. Method 1 presents the lowest variance.