MR flow measurements for assessment of the pulmonary, systemic and bronchosystemic circulation: Impact of different ECG gating methods and breathing schema

1. Introduction 

MR phase-contrast flow (PC-flow) measurements have been established to assess flow in large vessels [1], [2]. In phantom experiments a high accuracy of PC-flow measurements was found [3], [4]. PC-flow measurements are mainly based on 2D FLASH sequences with ECG triggering. This gating is most often done by a prospective analysis of the ECG cycle. The ECG gating unit continuously monitors the heart cycle and after detection of the R-wave the trigger signal is sent to the MR scanner. This introduces a small delay time leading to a reduced capability to assess the early systolic blood flow. In addition, the RR-interval time is manually adjusted with a 10% shortage of the detected cycle (arrhythmia rejection window). Therefore, the late diastolic blood flow is not measured. Putting both facts together, prospectively ECG gated measurements will underestimate blood flow by principle. To allow for more precise flow measurements retrospective ECG triggering was introduced in the early 1990s [5], [6]. However, it took the MR manufacturers several years to implement this technique in routine imaging sequences. Retrospective triggering was compared to prospective triggering for quantification of the left ventricular volume and ejection fraction comprising a detailed description of the gating technique [7]. However, up to now no study has evaluated the impact of the ECG gating technique on PC-flow measurements. It is important to demonstrate whether the results of examinations performed with different ECG gating techniques are comparable, especially for follow-up studies. Another point for daily practice is the time of acquisition needed for flow measurements. Free breathing acquisition take 3–4min, while breath-hold acquisitions can be performed in less than 30s. To achieve these short acquisition times, the temporal resolution has to be reduced for prospective gating, while using retrospective ECG gating the temporal resolution can be kept comparable to standard prospective flow measurements.

Therefore, the goal of our study was to evaluate the impact of a free breathing and an expiratory breath-hold retrospective ECG gated PC-flow measurement in comparison to a standard free breathing, prospectively ECG gated PC-flow measurement on quantitative flow evaluation of the pulmonary, systemic and bronchosystemic circulation.

2. Material and methods 

Seven female and 13 male healthy volunteers (mean age 25±5 years) were included in this prospective study. The mean height was 177±10cm and the mean weight was 71±11kg. All volunteers had no history of cardio-thoracic surgery, intracardiac shunts, tumours or arrhythmia. Ethics committee approval was obtained prior to investigation and all volunteers gave informed consent.

All examinations were performed on a 1.5T whole body MR system (Magnetom Avanto, Siemens Medical Solutions, Erlangen, Germany) equipped with the Syngo software version 2004V. For signal detection the manufacturer's body matrix coil and spine coil were used. No contrast media was injected. After acquisition of multidirectional localizers segmented 2D FLASH phase-contrast flow measurements were planned perpendicular to the pulmonary trunc and the ascending aorta. The velocity-encoding for flow measurements in the pulmonary trunc was 150 and 300cm/s in the ascending aorta.

In ascending aorta and the main pulmonary trunc three PC-flow measurements were performed: first, free breathing, prospective gating (free-prospective), second free breathing, retrospective gating (free-retrospective) and third expiratory breath-hold, retrospective gating (bh-retrospective).

3. Results 

All volunteers tolerated the examination well. For each volunteer six PC-flow datasets were acquired and evaluated. The results are summarized for the pulmonary trunc (Table 1) and for the ascending aorta (Table 2). One example of the assessed mean velocity in the ascending aorta is given in Fig. 1. The most important findings for clinical routine are presented below.

Table 1. Summary of the results of the different measurements in the pulmonary trunc

	Parameter	(1) Free breathing prospective ECG		(2) Free breathing retrospective ECG		(3) Expiratory breath-hold retrospective ECG
		Mean±S.D.	p-Value 1 vs. 2	Mean±S.D.	p-Value 2 vs. 3	Mean±S.D.	p-Value 1 vs. 3
	RR-interval [ms]	976±162	0.4	974±162	0.005	905±134	0.005
	Number of frames	31±5	<0.001	70±14	<0.001	28±4	0.002
	Temporal resolution [ms]	28±0.5	<0.001	10±2	<0.001	24±2	<0.001
	Difference in diastole [ms]	120±43	<0.001	21±5.5	<0.001	45±17	<0.001
	Vessel distensibility [cm2]	2.9±1.3	0.3	3±1.1	0.3	2.9±1.1	1
	Average flow [l/min]	6±1.4	0.003	6.4±1.6	0.01	6.9±1.6	<0.001
	Peak velocity [cm/s]	83±18	0.5	82±16	0.04	83±16	0.5
	Time to peak velocity [ms]	154±35	0.01	141±36	0.3	149±35	0.2
	Peak of mean velocity [cm/s]	60±13	0.09	61±14	0.1	62±13	0.006
	Time to peak of mean velocity [ms]	154±26	0.02	141±28	1	142±20	0.03

Data are given as mean±standard deviation (S.D.).

Table 2. Summary of the results of the different measurements in the ascending aorta

	Parameter	(1) Free breathing prospective ECG		(2) Free breathing retrospective ECG		(3) Expiratory breath-hold retrospective ECG
		Mean±S.D.	p-Values 1 vs. 2	Mean±S.D.	p-Values 2 vs. 3	Mean±S.D.	p-Values 1 vs. 3
	RR-interval [ms]	957±162	0.9	957±164	0.1	919±169	0.2
	Number of frames	30±6	<0.001	71±11	<0.001	28±5	0.003
	Temporal resolution [ms]	28±0.4	<0.001	10±0.6	<0.001	24±3	0.001
	Difference in diastole [ms]	108±33	<0.001	20±7	<0.001	38±11	<0.001
	Vessel distensibility [cm2]	1.6±0.4	0.01	2±0.6	0.02	1.6±0.5	0.5
	Average flow [l/min]	5.6±1.3	0.02	5.8±1.5	0.001	6.7±1.9	<0.001
	Peak velocity [cm/s]	112.8±21.4	0.03	109±23.1	0.07	113±25	0.5
	Time to peak velocity [ms]	139±28	0.001	125±29	0.6	122±27	0.003
	Peak of mean velocity [cm/s]	75±16	0.9	74±16	0.01	78±18	0.04
	Time to peak of mean velocity [ms]	129±21	0.02	116±22	0.3	111±17	0.005
	Time of image acquisition	2min 51s±29s		4min 36s±47s		17s (± 3s)

Data are given as mean±standard deviation (S.D.).

View Large Image Download to PowerPoint

Fig. 1. Mean velocity assessed in the ascending aorta in one male volunteer. Each of the three phase-contrast flow measurements techniques are displayed as separate curve.

3.2. Systemic circulation 

In the ascending aorta, the average blood flow per minute was 5.6l/min (corrected for coronary blood flow 5.8l/min) for free-prospective, while it was significantly higher (5.8l/min, p=0.02; corrected for coronary blood flow 6.1l/min) for free-retrospective and 6.7l/min (corrected for coronary blood flow 7.1l/min) for the bh-retrospective measurement (Fig. 2).

View Large Image Download to PowerPoint

Fig. 2. Net forward flow over time of one male volunteer showing the typical distribution of the curves. The net flow is largest in expiration, while the retrospective ECG gated free breathing measurement lies between the two others.

The peak of mean velocity was 75cm/s for the free-prospective, while it was 74cm/s (p=0.9) for free-retrospective and 78cm/s for the bh-retrospective (p=0.04 compared to free-prospective, p=0.01 compared to free-retrospective) measurement.

3.3. Bronchosystemic shunt 

The difference between the blood flow in the ascending aorta (corrected for the coronary arterial and bronchial arterial flow) and the pulmonary circulation was 0.12l/min for the free-prospective technique and 0.5l/min for the bh-retrospective measurements (p=0.04 between those two measurements). The difference was 0l/min for the free-retrospective measurement (Table 3).

Table 3. Calculation of the bronchosystemic blood flow

	Parameter	(1) Free breathing prospective ECG		(2) Free breathing retrospective ECG		(3) Expiratory breath-hold retrospective ECG
		Mean±S.D.	p-Values 1 vs. 2	Mean±S.D.	p-Values 2 vs. 3	Mean±S.D.	p-Values 1 vs. 3
	Average flow [l/min]+coronary blood flow (5%)	5.8±1.4	0.02	6.1±1.6	0.001	7.1±2	<0.001
	Systemic flow−pulmonary flow	−0.17±0.4	0.2	−0.3±0.4	0.06	0.15±0.9	0.2
	Average flow [l/min]+coronary blood flow (5%)+bronchial systemic flow (5%)	6.1±1.5		6.4±1.7		7.4±2.1
	Systemic flow (added bronchial flow)−pulmonary flow	0.12±0.4	0.2	0±0.4	0.1	0.5±1	0.04

First, the average blood flow in the ascending aorta [l/min] was increased by 5% to compensate for the coronary blood flow. The difference between the systemic blood flow and the pulmonary arterial blood flow indicates the bronchoystemic shunt. The normal value for this shunt volume is 5% of the cardiac output of the left ventricle. Therefore, another 5% were added to the systmic blood flow and again the difference was calculated.

4. Discussion 

Using retrospectively gated 2D FLASH flow measurements the temporal resolution was significantly higher than for prospective gating. Furthermore, the RR-interval was covered completely with no gap at early systole and only a marginal gap during late diastole. The measured blood flow per minute was significantly higher using retrospective ECG gated flow measurements (up to 0.9l/min compared to prospective ECG gating). The blood flow in the pulmonary circulation was lower (120ml/min) for the free-prospective, and also for the bh-retrospective measurements (500ml/min), while the difference was 0ml/min for the free-retrospective. This result emphasises the need for high temporal resolution for accurate measurements of hemodynamic using MRI.

The main drawback of a free breathing image acquisition is the long acquisition time, often exceeding several minutes. Therefore, breath-hold techniques were implemented in daily clinical routine, to allow for assessment of flow in 20–30s [10]. However, it is known that especially deep inspiration leads to a significant reduction in cardiac output due to a decreased venous return to the heart. The most important effect of breath-holding on cardiac performance is the difference in intrathoracic pressure [11]. Ferringno et al. measured cardiac output and intrathoracic pressure in six healthy volunteers using impedance cardiography, and observed a 24% decrease in cardiac index with breath-holding at large lung volumes due to increased intrathoracic pressure and decreased venous return. Therefore, in early studies it was recommended to perform flow measurements during free breathing in order to acquire precise flow information being least influenced by physiological variation [12]. In a later study it was shown, that blood flow quantification acquired at low lung volumes is close to physiological blood flow [13]. Therefore, the breath-hold examination in our study was done in expiration.

A technical drawback of the breath-hold techniques was the low temporal resolution of approximately 40ms [14], [15]. On the other hand it is recommended to acquire 30 frames during the RR-interval [3]. This high number of frames cannot be acquired using breath-hold prospective ECG gated flow measurements. By use of retrospective ECG gating it was possible to acquire 28 frames in one expiratory breath-hold (bh-retrospective). The temporal resolution using the bh-retrospective measurement was significantly higher than in free-prospective (24ms versus 28ms, p<0.001), although the time of acquisition was only 17s. However, in our study population heart rate and cardiac output during the expiratory breath-hold were significantly higher than during free breathing. This demonstrates that even a breath-hold of 17s already induces significant physiological hemodynamic changes. Thus, a breath-hold acquisition, even with high temporal resolution, is strongly discouraged.

Using retrospective gating techniques an interpolation of time frames based on the raw data is done with the drawback of averaging effects due to too many iterative interpolation steps [6]. Therefore, we set the number of calculated phases only twice as high as the TR of the sequence might have allowed for prospective gating.

Due to more complete coverage of the cardiac cycle the blood flow per minute was higher using retrospectively ECG gated measurements. Therefore, the hemodynamic parameters from prospective or retrospective ECG gating need to be compared with normal values determined with the same gating technique, and for follow-up examinations it is essential to use the same gating technique in order to be able to assess therapy effects.

There was only a minor impact of interpolation onto the pulsatile aspect of the flow characteristics. The absolute peak velocities in the ascending aorta ranged from 109 to 113cm/s (p=0.03–0.5) and between 82 and 83cm/s (p=0.04–0.5) in the pulmonary artery. Although some results were statistically significant, a maximum difference of 4cm/s between these methods may be not relevant in clinical routine.

If MRI is used for non-invasive estimation of pressure, the correct identification of the time to peak velocity is essential. In the pulmonary trunc it was shortest, 141ms, by using the highest temporal resolution (free-retrospective). Both other techniques showed a significantly longer time to the maximum peak velocity (149ms (bh-retrospective) and 154ms (free-prospective), respectively).

In previous studies there was a tendency for better correlations of MR derived parameters and pulmonary arterial pressure with higher temporal resolution [16], [17], [18]. A temporal resolution of 10ms was used in an animal model and showed excellent correlation with the invasively measured pulmonary arterial pressure [19]. Thus, for the purpose of non-invasive pressure estimates a retrospectively ECG gated free breathing technique should be used.

The lungs are supplied by a dual arterial system composed of the pulmonary arteries and the bronchial arteries (vasa vasorum) [9]. By means of MR flow measurements in the systemic and pulmonary circulation the bronchosystemic shunt volume can be calculated [10]. The shunt volume can be enlarged either in pulmonary arterial or lung parenchymal pathology [20]. This shunt can be used as a surrogate parameter for therapy control (like reduction of parenchymal inflammation) in follow-up examinations. In our study no difference between the flow in the systemic circulation (corrected for the coronary artery and bronchial flow) and the pulmonary circulation was found for the free-retrospective. Thus, for the assessment of the bronchial arterial blood flow a retrospectively ECG gated free breathing technique should be used.

A drawback of this study is that the only young volunteers with a regular heart rhythm were examined. In older patients with generally reduced cardiac output the differences between the gating strategies may not have been that large. Transforming these results to sick patients with faster heart beats than young volunteers, it will further reduce the number of phases which can be acquired with prospective gating. Therefore, the retrospective ECG gated technique will be even more important. As MR PC-flow measurements were considered the standard investigation, no dedicated phantom studies were performed during this investigation. The findings of this study increase the need for further physiological (animal) experiments with invasive measurements to evaluate the effect of an increasing number of time frames per RR-interval systematically.

5. Conclusion 

The type of ECG gating has a substantial impact on the absolute hemodynamic values determined by MR PC-flow measurements. Especially in follow-up examinations it is essential to use the same gating technique to be able to assess therapy effects. Only the free breathing, retrospectively ECG gated measurement allowed for assessment of the physiological bronchosystemic blood flow.

Expiratory breath-hold techniques lead to a significant change of cardiac and hemodynamic physiology and should not used in clinical routine.