Testing

Steps to ensuring safety of ECGs

The following 8 points are based on Design and Development of a Validation Framework for ECG Device Stabilisation. This list is informative, not official. Think of these as the "10 Commandments of ECG" (except there are 8).

1. Insulate users from voltage fluctuations or electrical shocks
2. ECG must detect signals of the heart only, and reject signals from other sources
3. Defibrillation/voltage surge protection
4. Preservation of ECG signal without interference from noise
5. Protection against environmental emissions
6. Able to detect all (7) types of arrhythmias
7. Able to detect pacemaker pulses (by oversampling)
8. Able to reject tall T wave signals beyond a limit to prevent heart beat duplication

Essential ECG Testing

Essential performance requirements are as follows:

1. Defibrillation Protection:

• Within 5 s after exposure to the defibrillation voltage of 5kV, the ECG shall resume normal operation in the previous operating mode, without loss of any operator settings or stored data, and shall continue to perform its intended function
• Two defibrillation tests need to be done, Common mode and Differential mode

2. Essential Performance and Accuracy of ECG machine

• Only needed if automated measurements are provided by the machine - refer to Annex AA.3 on pg.52 for guidelines to input ECG data to ECG

a) Requirements for amplitude measurements – refer to Table GG.1 on pg.63

• Exclude the two biggest differences in the amplitude measurements. The difference for each remaining amplitude measurement shall not deviate from the reference value by more than ±25 μV for reference values ≤500 μV, or by more than 5 % or ±40 μV (whichever is greater) for reference values >500 μV.

b) Requirements for absolute interval measurements – refer to Table GG.1 on pg 63

c) Requirements for biological interval measurements – refer to Table GG.2 on pg. 64

3. Filters (including line frequency interference filters)

• Filters for line frequency interference suppression shall not introduce on the ECG report more than 50 μV peak-to-valley distortion of the signal in any lead when tested with the test ECG ANE20000 (on pg. 63 of Table GG.1).

4. Electrostatic Discharge

• The machine shall resume normal operation in the previous operation mode without loss of operator settings or stored data within 10 s of a discharge (temporary degradation is allowable for up to 10 s).

5. Electrical fast transients and bursts

• When exposed to electrical fast transients and bursts (via the power cord supply), the machine shall resume normal operation in the previous operation mode without loss of operator settings or stored data within 10 s. The machine shall comply with the requirements of amplitude measurements when the signal CAL20110 of Table GG.1 is applied.

6. Conducted Disturbances

• When exposed to a conducted radio frequency voltage, via the power cord supply, the machine shall continue to perform its intended function. The machine shall comply with the requirements of amplitude measurements when the signal CAL20110 of Table GG.1 is applied. The difference for each amplitude measurement shall not deviate from the reference value by more than ± 50 μV for reference values ≤ 500 μV, or by more than 5% or ±100 μV (whichever is greater) for reference values > 500 μV.

7. Electrosurgery interference

• When the machine is used together with high frequency surgical equipment, it shall return to previous operating mode within 10 s after exposure to the field produced by the high frequency surgical equipment, without loss of any stored data.

Calibration Testing

CSE Test Atlas

Electrocardiograms currently on the market fall into two categories: traditional analogue ECGs capable of printing ECG reports on paper or displaying them on a screen, and more modern digital ECGs with integrated signal processing, capable of making measurements and interpretative statements on the ECG report. A major goal of the European CTS-ECG project was to develop signals that can be fed into both older analogue and newer digital ECGs, and therefore be used for both signal reproduction verification on paper-strip as well as for computer programs which have been designed to analyse and interpret ECG reports.

As a result a set of ECG-like signals with well-defined amplitude-time characteristics were created. The signals are generated in digital form and can be used to analyse both:

• The hardware characteristics of analogue systems in terms of amplifier linearity, gain factors, weighting factors for lead networks, low and high frequency response, and signal reproduction on printed reports;
• The software performance, for example in terms of waveform detection, recognition of fiducial points, measurement of ECG parameters, etc.

PhysioNet

One of the major drawbacks of the CSE study is that their database is only available on CD-ROM at high cost (ie closed-source). In an effort to improve transparency, many freely available databases have been provided within PhysioNet, an open source research resource for complex physiological signals including the ECG. The website contains a large cache of clinically validated ECG signals, and also houses open-source software programs that can be used to view, process, and analysis these complex signals. In keeping with Glia’s ideology of making medical grade equipment that is easily accessible, using PhysioNet may be the ideal toolkit to test and validate a prototype ECG machine. The use of databases provides a realistic range of data from actual patients, along with validated annotations from cardiologists. However, transitioning a digital signal to analog to test the front end hardware acquisition can present a number of challenges, such as inherent noise fluctuations, difficulty measuring clinical parameters of the ECG, and using any particular database may prevent some specific waveforms from being presented.

For a list of available ECG databases on PhysioNet, follow this link: https://physionet.org/physiobank/database/#ecg

Conversely, use of an artificial signal is noise free and contains known clinical parameters that can be controlled. By varying the model over all possible heart rates, leads, and rhythms, and measuring the difference in all the clinical parameters, it is possible to rapidly determine under what circumstances the acquisition hardware causes significant distortions in the clinical parameters measured from the ECG. Conveniently, PhysioNet is also home to a well-known open-source algorithm ECGSYN which can be used to calibrate biosignal acquisition devices such as the ECG machine.