Ischemia-related intra-QRS changes. When the heart muscle becomes ischemic or infarcted, characteristic changes are seen in the form of elevation or depression of the ST-segment. Detection of these changes requires an extension of the signal bandwidth to frequencies down to 0.05 Hz and less, making the measure- ments susceptible to motion artifact errors. Ischemia also causes changes in conduction velocity and action potential duration, which results in fragmentation in the depolarization front (Figure 5.3) and appearance of low-amplitude notches and slurs in the body surface ECG signals. These signal changes are detectable with various signal processing methods [33,34]. Depolarization abnormalities due to ischemia may also cause arrhythmogenic reentry [35], which is one more reason to detect intra-QRS changes precisely.
Identification of ischemia-related changes in the QRS complex is not as well known, and interpretation of the QRS complex would be less susceptible to artifactual errors, as compared to the ST analysis. Thus,
(a) Normal depolarization (b) Ischemic depolarization
Depolarization Depolarization
Ischemic zone Isochrones
Left ventricle Left ventricle
Observation points (electrode sites)
Observation points (electrode sites)
Site of LAD occlusion
FIGURE 5.3 Idealized example of (a) normal propagation and (b) wavefront fragmentation due to an ischemic zone of slow conduction. The superimposed lines are isochrones, connecting points at which the depolarization arrives at the same time.
time-frequency or time-scale analysis would serve a useful function in localizing the ischemia-related changes within the QRS complex, but would be somewhat independent of the artifactual errors.
Experimental findings. In experimental animals, the response of the heart to coronary artery occlusion and then reperfusion was studied. The Left Anterior Descending (LAD) branch of the coronary artery was temporarily occluded for 20 min. Subsequent to that, the occlusion was removed, and resulting reperfusion more or less restored the ECG signal after 20 min. The coronary artery was occluded a second time for 60 min, and once again occlusion was removed and blood flow was restored. Single ECG cycles were analyzed using the continuous wavelet transform [33]. Figure 5.4 shows the time-scale plots for the ECG cycles for each of the five stages of this experiment. The three-dimensional plots give time in the P-QRS-T complex on one axis, the scale (or equivalent frequency) on another axis, and the normalized magnitude on the third axis. First occlusion results in a localized alteration around 100 ms and the midscale, which shows up as a bump in the three-dimensional plot or a broadening in the contour plot. Upon reperfusion, the time-scale plot returns to the pre-occlusion state. The second occlusion brings about a far more significant change in the time-scale plot, with increased response in the 0 to 200 msec and mid-scale ranges. This change is reversible. We were thus able to show, using time-scale technique, ischemia related changes in the QRS complex, and the effects of occlusion as well as reperfusion.
Potential role of ischemia-related intra-QRS changes in coronary angioplasty. The above results are also applicable to human ECGs and clinical cardiology. For example, a fairly common disorder is the occlusion of coronary vessels, causing cardiac ischemia and eventually infarction. An effective approach to the treatment of the occlusion injury is to open the coronary blood vessels using a procedure called coronary angioplasty (also known as percutaneous transluminal coronary angioplasty or PTCA). Vessels may be opened using a balloon-type or a laser-based catheter. When reperfusion occurs following the restoration of the blood flow, initially a reperfusion injury is known to occur (which sometimes leads to arrhythmias) [35]. The ST level changes as well, but its detection is not easy due to artifacts, common in a PTCA setting. In a clinical study, we analyzed ischemia and reperfusion changes before and after the PTCA procedure. Short-term occlusion and ischemia followed by reperfusion were carried out in a cardiac catheterization laboratory at the Johns Hopkins Hospital in connection with PTCA) [36].
Figure 5.5 shows time-scale plots of a patient derived from continuous wavelet transform. Characteristic midscale hump in the early stages of the QRS cycle is seen in the three-dimensional time-scale plot.
Then, 60 min after angioplasty, the normal looking time-scale plot of the QRS complex is restored in this patient. This study suggests that time-scale analysis and resulting three-dimensional or contour plots may be usable in monitoring the effects of ischemia and reperfusion in experimental or clinical studies. In another study (4 patients, LAD) we monitored for intra-QRS changes during PTCA. Despite signal noise and availability of recordings only from limb leads, superimposed mid-frequency components during ischemic states of the heart were observed, which disappeared when perfusion was restored. There was at least one lead that responded to changes in coronary perfusion. Figure 5.6 shows five different stages of a PTCA procedure as time plots (lead I) and CWT TFDs (topo-plots). Despite the presence of noise, the WT was able to unveil elevation of intra-QRS time-frequency components around 20 Hz during balloon inflation (ischemia), and a drop in the same components with reperfusion after balloon deflation.
Frequency components 20 to 25 Hz during inflation. No substantial ST changes can be observed in the time-domain plot. The arrows show the zones of change in TFDs with ischemia and reperfusion. Note the representation of power line interference (50 Hz) as“clouds”in (b) and (d) topo plots — far from the region of interest.
Another study analyzed ECG waveforms from patients undergoing the PTCA procedure by the mul- tiresolution wavelet method, decomposing the whole P-QRS-T intervals into coarse and detail components [26,37], as can be seen from the analysis of one pre-angioplasty ECG cycle in Figure 5.7 [26]. The PTCA procedure results in significant morphological and spectral changes within the QRS complex. It was found that certain detail components are more sensitive than others: in this study, the detail components d6 and d5 corresponding to frequency band of 2.2 to 8.3 Hz are most sensitive to ECG changes following a successful PTCA procedure. From this study it was concluded that monitoring the energy of ECG signals at different detail levels may be useful in assessing the efficacy of angioplasty procedures [37]. A benefit of
Panel A Panel B Lead II normal perfusion20 min reperfusion20 min occlusion20 min reperfusion20 min occlusion
CWT normalized magnitude →
(a)(b)(c)(d)(e)
400 200 0
400 200 0
400 200 0
400 200 0
400 200 0
16 200 Hz
(log) 16
200 Hz
(log) 16
200 Hz
(log) 16
200 Hz
(log) 16
200 Hz
(log) Time [msec] ←
← Scale frequency →← Scale frequency →← Scale frequency →← Scale frequency →← Scale frequency →
← Time [msec]
← Time [msec]
16(log)200 Hz16(log)200 Hz16(log)200 Hz16(log)200 Hz16(log)200 Hz
400 200 0
400 200 0
400 200 0
400 200 0
400 200 0
FIGURE5.4Time–frequencydistributionsofthevectormagnitudeoftwoECGleadsduringfivestagesofacontrolledanimalexperiment.Thefrequencyscaleislogarithmic,16to 200Hz.Thez-axisrepresentsthemodulus(normalized)ofthecomplexwavelet-transformedsignal.
(b)
CWT magnitude CWT magnitude
16 Hz (log)
200 Hz
←Scale
frequency
→
16 Hz (log)
200 Hz
←Scale
frequency
→
← Time [msec]
← Time [msec]
400 300
200 100
0
400 300
200 100
0
Pre-PTCA 60 min after PTCA
(a) (b)
FIGURE 5.5 Time–frequency distributions of human ECG study using WT. Pre-angioplasty plot (a) shows a char- acteristic hump at about 35 Hz, which disappears as indicated by the second plot (b) taken 1 h after angioplasty treatment.
this approach is that a real-time monitoring instrument for the cardiac catheterization laboratory can be envisioned (whereas currently x-ray fluroscopy is needed).
Detection of reperfusion during thrombolytic therapy. Detecting reperfusion-related intra-QRS changes, along with ST changes, in the time–frequency domain would possibly find application in thrombolysis monitoring after myocardial infarction. At present, the ST elevation and its recovery are the main electrocardiographic indicators of acute coronary ischemia and reperfusion. Reports using continuous ST-segment monitoring have indicated that 25–50% of patients treated with intravenous thrombolytic therapy display unstable ST recovery [38], and additional reperfusion indicators are necessary. Signal averaged ECG and highest frequency ECG components (150–250 Hz) have been utilized as a reperfusion marker during thrombolysis and after angioplasty, but their utility is uncertain, since the degree of change of the energy values chosen does not appear to be satisfactory [39,40]. We have analyzed the QRS of the vector magnitude V =of body surface orthogonal ECG leads X , Y and Z during thrombolytic therapy of two patients with myocardial infarction. Figure 5.8 shows how TFDs may be affected by reperfusion during thrombolysis. Two interesting trends may be observed on this figure: (1) a mid-frequency peak present during initial ischemia (a) disappears two hours after start of thrombolytic therapy (b) due to smoother depolarization front, and (2) high-frequency components appear with reestablished perfusion, possibly due to faster propagation velocity of the depolarization front.