Rejection Using a VCG-Based Triggering Algorithm

JM Chia’, SE Fischer’,’ , SA Wickline,‘, CH Lore& ICenter for Cardiovascular MR, Cardiovascular Division, Washington University Medical Center, St. Louis MO, USA 2Philips Medical Systems, Best, The Netherlands

Introduction With standard QRS detection algorithms, arrhythmic heartbeats can be mistaken as normal QRS complexs. This is undesirable for triggered cardiac magnetic resonance examinations since image quality will be lowered when the R-R duration of heartbeats preceding and following ectopic beats varies. Current MR arrhythmia rejection methods compare each R-R interval to a reference interval in order to accept or reject retrospective data associated with each trigger This method fails in certain types of and is susceptible to rate variation, i.e. during a cardiac test It has been shown using the vectorcardiogram (VCG) extra-systolic QRS loops differ in shape and orientation Figure 2 -Frontal projection of volunteer with PVC from normal QRS loops [1] Using this information, we propose that by using a VCG-based target-distance triggering algorithm Figure 2 shows the frontal VCG plot from a volunteer. Here it can [2], certain types of arrhythmias can be rejected. clearly be seen that the PVC has a different shape than the QRS loop. In addition, the PVC is in a different orientation, so it does not Methods pass through the reference point. Thus, the VCG target-distance A total of six patients were studied. In this group there were five algorithm does not mistake the PVC for a normal QRS loop males and one female with an average age if 63i18 Five of the volunteers had premature ventricular contractions (PVC) while IID #I Arrhvthmic ORS 1 Normal ORS 1 one patient had supraventricular extra-systoles They underwent Type Rejected Total % Rejected cardiac functional examinations on a Philips Gyroscan 1 5T whole-body system triggered with the VCG triggering system developed in our lab using a target-distance algorithm [2] On ~1 each patient two pairs of electrodes were placed over the heart in a cross orientation to achieve ECG projections close to x and y orthogonal leads of the VCG During the MR exam the ECG 1 5 1 PVC 1 206 1 206 I 100 I 2702 I 99.98 1 waveforms and trigger signal were acquired at a 50 Hz sampling Table I - Summary of results (PI = Performance Index, SVES = rate and saved After the MR scan the arrhythmic and normal QRS Supraventricular Extra-Systole) complexes were identified semi-automatically and compared to the trigger signal using a custom developed software package. Table 1 shows the summary of results for the 5 volunteers. In three of the volunteers, all arrhythmic QRS complexes were Results and Discussion rejected. In all cases, the arrhythmias were PVC’s in which the loop In five of the six patients, a precise reference point was was in a different orientation than the QRS loop. In volunteer 4, no established to trigger the examination. However, in the sixth arrhythmias were rejected. This volunteer had supraventricular patient, due to the excessive frequency of PVC’s in the heart extra-systoles. In this type of arrhythmia, the arrhythmic loop is the rhythm, an accurate reference point could not be determined same shape and orientation as the normal QRS loop. Hence, the Thus, this patient was excluded from further analysis. algorithm triggers on the arrhythmic beats. The only distinguishing feature between this arrhythmia and the normal QRS loop is the x [mVl variation in the R-R duration Volunteer 3 also had PVC’s; however, not all the PVCs were rejected by the algorithm In those beats not rejected, although the PVC loop was a different shape and Y [mvl orientation than the QRS loop, the loop still passed through the established reference point of the algorithm, resulting in a false trigger. Even though not all arrhythmic QRS complexes were rejected, the performance index [3] of the VCG algorithm for normal QRS complexes was still very accurate (Table 1)

Figure 1 - Output signals Conclusions The potential for the VCG based target-distance algorithm to Figure 1 shows sample output signals from the VCG system. perform robust rejection of PVC’s has been demonstrated. With this Here it can be seen that the VCG algorithm can distinguish the consistent arrhythmia rejection, image quality should be improved. difference between a premature ventricular contraction (PVC) and However, more work will be needed to define effective rejection a normal QRS complex. Examinations of the ECG signals reveals schemes for other types of arrhythmias. that the PVC manifests only slightly different characteristics from those of the normal QRS complex. The differences are apparent in References the timing and the amplitude of the PVC. Here, it can be seen that 1 Talbot, S et al , British Heart Journal, 40, 883-890 (1978) the PVC occurs more closely in time after a normal QRS complex 2. Fischer, SE et al, MRM, 42,361-370 (1999) and thus, may be rejected by standard algorithms only if the R-R 3. Friesen, GM et al , IEEE Trans. Biomed. Eng, 37,85 (1990) window is set large enough to exclude it

Proc. Intl. Sot. Mag. Reson. Med. 8 (2000) 201