Physiological basis of electrocardiography

This chapter explores the relationship of Sections G2(iii) and G7(i) of the 2017 CICM Primary Syllabus, which ask the exam candidate to "explain the physiological basis of the electrocardiograph" and "describe the principles behind the electrocardiogram (ECG)". Though on superficial inspection this duplication of syllabus items would seem like evidence of design by committee, the attentive pedant could point out that an electrocardiogram is a graph of cardiac voltage plotted versus time, whereas an electrocardiograph is the device used to record it. More murky is the distinction between "physiological basis" and "principles", but hopefully this chapter is a sufficiently thorough survey of this topic, which should cover every possible meaning of those terms.

This has appeared multiple times in the past papers, offering a detailed view of examiner expectations. 

• Question 17 from the second paper of 2021
• Question 9 from the first paper of 2016
• Question 5 from the second paper of 2011
• Question 13 from the first paper of 2011
• Question 1 from the first paper of 2009

The pass rate for some of these ranged as low as 8%, making this topic something of a revision priority. "Future candidates need to be aware that such questions WILL get asked again", the examiners threatened. Overall, they seem to have wanted three things:

• Relationship of the surface ECG to the events of the cardiac cycle
• Essential components of an ECG monitor
• Methods used to decrease artifact and interferences 

In summary:

 

An absolute orgy of peer-reviewed literature awaits the trainee who is fed up with unreliable online resources. This is not exactly an exotic niche topic. Over 150,000 results appear for the Google Scholar search "physiological basis of the electrocardiogram", of which the topmost are literally titled "Physiological basis of the electrocardiogram". It is rather difficult to identify and recommend any specific article because there are several excellent ones, and they all cover roughly the same material in roughly the same way. One standout example is Chapter 2 from the ECG book by de Luna (2014), which contains a lot of highly satisfying detail written very accessibly, to please the enthusiast without deterring the late-night cram student. Reisner Clifford and Mark (2007) are another representative example which happens to be brief and freely available at the time of writing. This 37-page Bachelors thesis by Watanwall & Ekranian is also freely available, but not brief, and rendered unnecessarily difficult to interpret by some interesting translation errors. Lastly, if one really needs to intimidate their enemies with the size of their resource, Macfarlane's Comprehensive Electrocardiology (2010) covers this subject over 2291 pages. 

Definitions

Under most circumstances, professional literature seems to define the term "electrocardiogram" as "a graphical recording of the cardiac cycle produced by an electrocardiograph". From Mike Cadogan's LITFL review of ECG history, it appears that the origin of the term is Willem Enthoven who first introduced it as electro-cardiogrammem in 1895, and then gave us the PQRST labelling based (it appears) on Descartes' choice of letters for points describing a curve. Hurst (1998) looked into this and decided that the middle of the alphabet was chosen "to allow space for possible later additions". 

Relationship of surface ECG to the events of the cardiac cycle

As this is covered in great detail in the chapter dealing with events of the cardiac cycle, only some minimal summary of the main points will be mentioned here, to simplify revision:

• The SA node fires well into late diastole, and this is not represented on the surface ECG, nor is the propagation of signal along the internodal tracts.
• The P wave is produced as the atrial muscle depolarises. The right atrium contracts first.
• At the end of the P wave, the left atrium finally contracts. The end of diastole occurs during the PR interval.
• The R wave is generated by ventricular depolarisation, and its peak corresponds to the beginning of systole (specifically, of isovolumetric contraction). 
• The T-wave represents ventricular repolarisation, and corresponds to the phase of decreased contraction (slow ejection).  The peak of the T-wave correlates reasonably well with the onset of diastole, i.e the closure of the aortic valve.

Now, how do these ionic events actually create the familiar ECG trace? Our electrodes are attached to the patient's skin, not to the surface of their cells.  How can cardiac electrical activity be measured by external electrodes?

Observe.

Relationship of surface ECG to ionic events in the cardiac myocyte

The membrane of a cardiac myocyte has a negative internal charge, such that the charge difference between the inside and the outside of the membrane is about 90 mV, as shown here.

As the cardiac action potential propagates through the cell, the membrane is depolarised and the intracellular part becomes positive.  At the same time, the extracellular part becomes negative (partly owing to the loss of positively charged ions which end up being sucked across the membrane and into the myocyte).

So now, at this instant, we have a myocyte which has one positive end and one negative end, with the membrane action potential propagating towards the positively charged myocyte surface, leaving behind a negatively charged surface. In other words, at the interface between resting and depolarised cell membrane, there is a difference in electrical charge. This charge difference between two neighbouring intracellular regions of the myocyte produces an electric field, which is conventionally depicted as lines going radially outward from a positive charge and radially in toward a negative charge:

Of course in the depolarising myocyte, at the site of Phase 0 of the cardiac action potential, the positively charged membrane is separated from the negatively charged membrane by an incredibly small distance:

This situation is described as a dipole moment. The myocyte behaves like an electric dipole, which is something of a mathematical abstraction. A dipole, strictly speaking, is some arrangement of two infinitely small points, one of positive and one of negative charge, which are separated by a small distance (for some mathematical purposes, an infinitely small distance). This dipole has a magnitude (equal to the strength of each charge multiplied by the separation of these charges) and it has a direction (which is the vector line joining the charged points).

At this stage, most resources (here, here, here and here) are guilty of a maddening omission. A dipole vector by convention points from the negative charge to the positive charge. But the inside of the cell membrane is what is usually measured by physiologists, and this changes from negative (-90 mV) to positive (+30-50 mV) as the action potential hits.  That would mean that the dipole moment vector points in the opposite direction to the direction of the action potential propagation. However, most textbooks tend to represent the direction of propagation of the action potential as moving towards a positively charged region, leaving behind negatively charged depolarised cells. That is because they are describing the extracellular dipole:

Extracellular charge changes from positive to negative. Thus, looking at the changes in the surface charge of the cardiac myocytes, the direction of the dipole vector is the same as the direction of the action potential propagation: towards resting cells with positive surface charge, and away from depolarised cells with negative surface charge.

Geselowitz (1964) describes this in great detail, and it would be rather impossible to outdo his explanatory prowess, other than to depict what happens when myocytes in a neatly carved cube of myocardium all depolarise together. Observe: the white glowy part of this meat cube represents the site of the action potential, where every cell's membrane is in Phase 0.

Changs in total cardiac dipole moment during the cardiac cycle

Following from all this, one can behold the myocardium, a topographically complex three-dimensional object, as a large group of individual electric dipoles. It does not help that some of these dipoles are pointing in completely different directions. Fortunately, it does not matter too much. Looking at it from a large enough distance, each individual dipole becomes irrelevant and the whole myocardium can be considered a single dipole, as in slightly modified diagram from Holt et al (1969):

The sum of all their magnitudes and vector directions produces a net vector and a net magnitude. Thus, the whole of the myocardium can be described as a single electric dipole, which changes magnitude and direction as the wave of action potentials propagates across it. As the depolarisation affects different parts of this structure at different times, over the course of a single heartbeat the direction and magnitude of this summed dipole changes. One could represent these parameters as a diagram of an arc trajectory, drawn by the tip of the cardiac dipole vector as it moves over the course of a cardiac cycle. This sort of diagram is present in most textbooks; it was reconstructed from several sources (mainly Hobbie et al, 2015, and Hobbie, 1972).

As this dipole generates an electric field, it also changes its properties as the dipole moves around during the cardiac cycle. This changing electric field forms the basis of the ECG.

Relationship of the cardiac dipole to the body surface voltage 

As already mentioned above, the heart, being a huge electric dipole, produces an electric field. This field extends in all directions, and can be measured at the body surface.

Taking things way back to high school physics, one might recall that work is required to move a charge inside an electric field. When one tries to move a charge between two areas within an electric field, one needs to do work, and that work is described as the electrical potential difference, or voltage. Along the map of an electric field, there will be areas where the amount of work required will be the same, and these are usually described using "equipotential" or "isopotential" lines. It appears that the first recording of these equipotential lines on the surface of the human body was made by Augustus Waller in 1887. Here is a surviving version of the original which has been unnecessarily annotated for no apparent reason:

Measuring the potential difference between from two points along the same equipotential line will yield a potential difference of zero. Conversely, putting electrodes at two different points which each have a different electric field strength will produce a potential difference of something not zero.

From this, one can intuitively see the connection to the electrocardiograph. If you connect electrodes to two different points at the body surface, you should be able to measure the potential difference generated by the cardiac electric field, and then plot its changes over time as the cardiac dipole vector does its crazy dance during each cardiac cycle. Voila: the ECG.

Electric field measurement at the body surface

It is at this point that Einthoven's triangle is usually trotted out. Its base is Lead I, connecting the patient's arms. To illustrate on this Vitruvian man:

The negative electrode is conventionally on the right arm, and the positive electrode is on the left. As the result, it measures the potential difference produced by the difference in the electric field strength between the two arms. Observe: when the vector of the cardiac dipole is maximal in magnitude (at the peak of the R wave), the voltmeter connected to the arms would measure a positive deflection (i.e. a potential difference of +something volts). 

This positive deflection arises from the fact that the electric field produced by the depolarising myocardium is stronger in the left arm than in the right arm, and therefore the potential difference between electrodes (left arm minus right arm) produces a positive potential difference value.

So that deals with the magnitude of the cardiac dipole vector. But it also has a direction. If this direction was pointing directly at the left arm lead, the deflection in Lead I would be maximal; but because it points anteriorly and inferiorly, the magnitude of the vector is decreased. If additional electrodes are included in the measurement process, the direction of the vector can be triangulated. This is Einthoven's triangle.  As you can see from these excellent images (Dupre et al, 2005), the changes in vector direction is represented as the difference in ECG waveform height and direction between the different leads:

Thus, each pair of electrodes produces a potential difference, which is usually described as a "lead". At this point the textbooks typically digress on the topic of vector analysis and the additional precordial leads, but this does not appear to be essential for CICM exam preparation, and was deemphasised in the past papers, so we won't go there. For those who want to know more, Vieau & Iaizzo (2015) give a decent breakdown of how the other lead voltages are determined. From the perspective of the pragmatic exam-oriented trainee, the only thing that needs to be added is a discussion of the difference between bipolar and unipolar leads, of which "better answers made mention"

Leads I, II and III are bipolar electrodes. That is to say, they record the difference between two "poles", one positive and one negative, which are in anatomically different locations. "Unipolar" electrodes also have two poles, but the negative pole is not a real anatomical position per se, but rather an "indifferent" electrode made from the combined average potential difference measurements in the limb electrodes. It is also referred to as the Wilson electrode, or Wilson's Central Terminal, aftr F.N. Wilson who described it in 1934. Unfortunately nobody ever references Wilsons original article, because for some reason Malmivuo and R. Plonsey's Bioelectromagnetism is somehow more authentic and authoritative. Anyway, the point of it is to be a"neutral" electrode, which undergoes no change with cardiac activity, and therefore represents the central point around which the vector of the cardiac dipole revolves. The use of this artificially constructed reference point is to allow the measurement of the cardiac surface potential difference in a different plane, which is represented by the chest leads V1-V6.

Practical aspects of electrocardiograph design and function

The electric field strength, and therefore the potential difference, at the skin surface is rather poor. Abdul Jamil et al (2017) gives a range of 0.5-2.0 mV as the average potential difference in the limb leads. The maximal frequency content of this weak signal is said to be about 125 Hz. As the result, the recording of this potential difference and its fluctuations is subject to a lot of interference from a variety of sources. Thus, in order for the ECG signal to be interpretable, it needs to undergo some substantial modification. The steps involved in this are listed in the first few pages of the massive and exhaustingly precise AHA policy document (Kliegfield et al, 2007) which describes the design specification standards for ECG machines. What follows is a brief summary of those pages:

• Patient-machine interface: that is the flat sticky electrode attached to the patient's skin. The CICM answer to one of the ECG questions describes the electrodes as being made of "silver or silver chloride components", which seems excessively ostentatious for an impoverished public health system, but is in fact accurate. There's not a lot of precious metal in them.  Grassini et al (2013) describe these 10mm discs as "cheap and easily available". Their main design characteristic is low resistance, which is why they are flat and made from a highly conductive material.
• Signal acquisition elements: the potential difference needs to be transmitted to some kind of electrode, conveyed along some sort of conducting wire, and measured by some sort of galvanometer. Old-school ECG machines (up until the 1970s) used to collect analog signal, whereas these days everything is digital, with the conversion of analog signal usually happening at the level of the lead cable module (i.e. that chunky lead plug). The sampling rate for this digital signal is quite high (usually 10,000-15,000 Hz), not for the purposes of picking up higher-frequency cardiac signals but because most pacemaker pulse generators produce a very short (<0.5 msec) pulse, which would go unnoticed with a more relaxed sampling rate.
• Filtering: unwanted high-frequency noise is filtered out by software and hardware post-processing. There is in fact a lot of it. Here, a diagram presents some common frequency ranges for biologically generated electrical signals which the ECG electrodes end up measuring:
  
  
  The fundamental frequency for the QRS complex is around 10 Hz and according to Kliegfield et al (2007) most of the diagnostic information is contained in the sub-100Hz range, which means that you can safely discard any signals with a higher frequency. Generally, ECG machines filter the ECG to a range above 150 Hz, which tends to produce a civilised-looking smooth signal. Here's a nice comparison of a filtered and unfiltered trace from Bosznai et al (2009)
  
  
  Most of the high frequency signals are eliminated by this filter, leaving behind only pacing spikes. Additionally, low-frequency filtering also occurs, to eliminate the "baseline wander" associated with respiration. The low-frequency cutoff is usually 0.05 Hz; anything higher than this seems to produce spurious ST segment changes. 
• Amplification: most ECG machines have a gain of about 1000. As mentioned, the voltage of the signal is very low, and the ECG machine generally amplifies it significantly. The specific technique is called "differential amplification", owing to the fact that the ECG machine uses a differential amplifier to reinforce the difference between limb electrode input voltages. Since only the difference between the electrodes is amplified, any signal change which is common to both electrodes disappears. This is the basis for common-mode rejection.
• Common-mode rejection: this is a term used to describe the use of differential amplification to remove a signal which is detected more or less equally by all the measured leads. One such signal is the constant 50-60 Hz interference produced by all electrical mains wiring, casually referred to as "mains hum".
• Isolation: isolation amplifiers galvanically isolate both the input channel and the current-loop output driver from the controller hardware, preventing current leakage (Sommerville, 1990). ECG designers incorporate isolation to prevent component damage and to reduce mains interference (isolation prevents inductive-loop coupling between the signal transducer and the magnetic field generated by the 60Hz AC current in the mains-powered control circuit).
• Sufficiently high input impedance: A voltage amplifier with high input impedance will allow low-frequency current to pass with less resistance, excluding high-frequency noise but keeping the interesting low-frequency ECG signal.
• Earthing the ECG machine: current leakage from poorly earthed machines can give rise to artifact, which is usually not a problem as all patient-connected machinery is usually earthed. As such, these circumstances are rare enough to be confined to case reports about weirdly faulty modules, such as Selvan et al (2012).

It is possible to elaborate much more on this, but it is probably not essential from the ICU exam perspective. In case the reader is abnormally fascinated by electrical engineering, or somehow finds themselves possessed with the desire to design and build an ECG machine from scratch, Bailey et al (1990) contains the detailed breakdown of design recommendations for echocardiograph design, including frequency and phase filtering, fidelity criteria, data compression and storage, etc etc etc. 

None of that is essential for a ten-minute exam answer. From college comments, it would appear that the examiners mainly wanted some mention of the common causes of poor ECG signal and sources of interference, along with a brief outline of how these can be minimised.

Causes of poor ECG signal or artifact, and solutions to them

There are numerous possible causes for "artifact" in the ECG, which can be loosely defined as "any signal appearing on the ECG recording which does not originate from the myocardium". These can be grouped into three broad categories:

• Poor skin conductivity is a problem when the skin is dry, oily, or covered in thick coarse hair which puts air gaps under the electrode and decreases conductivity. Various conductive gels, goops and pastes come with electrodes and are available separately in tubs and tubes. In scenarios where it is for some reason impractical to smear the patient in this stuff, one resorts to shaving irregular patterns into their chest hair. This is not without risk, even aside from the adverse effects on manscaping; the shaver leaves behind tiny fragments of hair which can become incorporated into subsequent chest wounds, and damages hair follicles which can become infected (particularly when razors are used).
• Motion artefacts and EMG: the heart is not the only tissue in the body undergoing depolarisation; the other bulky tissues which generate electric fields are the nerves and muscles. Thus, muscle activity can be a source of interference. Additionally, by changing the position of lead electrodes, patient body movement (shivering, seizure, high-frequency oscillation, etc) can introduce changes into the ECG signal. The solutions present as sensible recommendations irrespective of the ECG recording context: rewarm the patient, ask them to stop moving erratically, actually take steps to manage the seizure instead of just standing there stupidly with the ECG machine, and stop using discredited ventilator technology. 
• Equipment-related electrical interference: both normal and faulty function of medical equipment can produce interference in the ECG leads. Patel et al (2008) list a whole range of possible culprits, including pumps, dialysis machines, ventilators, analog cell phones, diathermy equipment, orthopaedic power tools, and basically anything with defective insulation. Modern ECG machines come equipped with all manner of post-processing algorithms capable of removing some or all of this noise, and it helps that all electrodes tend to "hear" the same noise signal frequency (making it susceptible ot common-mode rejection) but sometimes "hum" penetrates these defences. The natural reaction to such problems would be to remove the faulty unit or pump, or at least pause its operation for the duration of the recording.

Diagnostic and monitoring modes

As this had come up in one of the previous college examiner comments, it is worth mentioning as an appendix. In essence, most ECG monitors and recorders have a continuous monitoring mode and a recording diagnostic mode, which differ mainly in the way they handle signal filtering. Switching to a "monitoring" mode implies that you are mainly interested in the rhythm, and so the machine focuses on removing environmental artifact. The frequency range of interest is therefore 0.5-40 Hz, eliminating high-frequency noise such as EMG produced by patient movement and nearby electrical devices. "Diagnostic" mode implies that you are interested in recording a detailed high-fidelity signal with a greater "resolution", and will take your own steps to remove artifact interference. In order to permit a more detailed analysis of ST segments and QRS morphology, the machine increases the range of permitted frequencies to 0.05-100 Hz.