Quantitative Assessment of Functional Variability with Real-time MRI
Markus Huellebrand1, Mathias Neugebauer1, Michael Steinmetz2, Jens Frahm3, and Anja Hennemuth1

1Fraunhofer MEVIS, Bremen, Germany, 2Universitätsmedizin Göttingen, Göttingen, Germany, 3Biomedizinische NMR Forschungs GmbH am Max-Planck-Institut für Biophysikalische Chemie, Göttingen, Germany


New real-time MRI imaging techniques enable the acquisition of serial images with a temporal resolution of up to 20 ms. These new imaging sequences provide cardiac parameters such as cardiac function and blood flow over multiple heart cycles and their variation over time. To quantify these parameters new analysis methods are required. Our proposed solution combines automatic image processing with interactive exploration techniques in a web application setup. The solution has been successfully tested with data from arrhythmic patients as well as volunteers performing Valsalva maneuver and physical exercise tests.


ECG-gated cardiac cine MRI is the gold standard for the assessment of cardiac function and blood flow. To overcome the limitations of conventional cine MRI in patients suffering from arrhythmia and shortness of breath, recent developments in acquisition technology provide real-time (rt) MRI sequences. This new technique allows for the acquisition of serial images with a temporal resolution of up to 20 ms under free breathing without ECG gating [1]. Recent studies of rt-MRI datasets have shown that single heartbeats as well as the variation of the cardiac contraction over time can be inspected quantitatively [2]. However, the application of conventional analysis tools is not possible because of the different image characteristics and the number of frames to process. The functionality to assess the additional information about beat-to-beat variability of clinical parameters is generally missing. The purpose of this work was the development of a software solution which enables quantitative analysis of rt-MRI data of function and flow in a clinical environment.


The proposed solution combines automatic image processing with interactive exploration techniques as shown in Figure 1. Data sent by the scanner is immediately pre-processed in a background process to analyze motion, detect heart cycles, and segment the myocardium in short-axis sequences [2,3]. The segmentation result as well as the temporal division into cardiac cycles can be corrected interactively. The exploration step is organized in such a way that analysis results can be inspected in a coarse-to-fine concept via a web browser. The slices as well as the heart cycles can be explored separately. The data to include for the calculation of overall functional parameters such as blood flow or ejection fraction adapts to the user selection. Thereby parameters can be calculated for different phases of the image series, e.g. for normal and ectopic beats separately.


The installation has been tested with data from arrhythmic patients as well as volunteers performing Valsalva maneuver and physical exercise tests. The size of the complete datasets was between 0.5 and 6 GB. Depending on the workload of the server (i7, 2.7GHz, 32GB RAM), automatic image analysis took between 1 and 30 minutes per image series. The effort for interactive correction was highest (10 min) for the exercise cases with strong through-plane motion affecting the image quality of the cine sequence. Figure 2 shows the comparison of the myocardial function analysis under physical stress with a pedometer and under rest (one section). The overall ejection fraction rose from 0.77±0.0 during rest to 0.84±0.03 during exercise. The multi-cycle analysis of the cardiac function of a patient with arrhythmia showed that the average ejection fraction of the irregular cycles was 0.43±0.02, whereas the normal EF was 0.66±0.04. The blood pool volume as a function of time in Figure 3 also clearly despicts the temporal shortening of the irregular heart cycles. Figure 4 shows a multi-cycle analysis of the blood flow in the ascending aorta from real-time velocity-encoded PC-MRI during a Valsalva maneuver. In the period during the maneuver the flow rate dropped from 90.04±4.98 ml to 63.11±10.12 ml, while the maximum peak velocity decreased from 80.42±12.75 cm/s to 57.23±8.26 cm/s.


We have developed a concept for the time-efficient analysis of rt-MRI datasets of cardiac function and flow. The solution has been successfully applied to compare cardiac blood flow and function under different stress levels, to measure the influence of breathing maneuvers such as Valsalva as well as to determine the influence of arrhythmia on cardiac function. The initial tests showed that cardiac variability can be quantified with the suggested acquisition and analysis pipeline with reasonable temporal effort. Future work will focus on studies to prove the clinical benefit of the provided technology.


No acknowledgement found.


1. Zhang S, Joseph AA, Voit D, Schaetz S, Merboldt KD, Unterberg-Buchwald C, Hennemuth A, Lotz J, Frahm J. Real-time magnetic resonance imaging of cardiac function and flow-recent progress. Quant Imaging Med Surg. 2014 Oct;4(5):313-29

2. Contijoch F, Witschey WR, Rogers K, Rears H, Hansen M, Yushkevich P, Gorman J 3rd, Gorman RC, Han Y. User-initialized active contour segmentation and golden-angle real-time cardiovascular magnetic resonance enable accurate assessment of LV function in patients with sinus rhythm and arrhythmias. J Cardiovasc Magn Reson. 2015 May 21;17:37.

3. Chitiboi T, Hennemuth A, Tautz L, Hüllebrand M, Frahm J, Linsen L, et al. (2014). Context-Based Segmentation and Analysis of Multi-Cycle Real-Time Cardiac MRI. In IEEE International Symposium on Biomedical Imaging (pp. 943–946).

4. Huellebrand M, Hennemuth A, Tautz L, Joseph A, & Frahm J (2015). Analysis of aortic blood-flow from ECG-free realtime PC MRI. In Journal of Cardiovascular Magnetic Resonance (Vol. 17 (Suppl 1), Q 42).


Figure 1: Concept of the server-based processing solution. Data sent from the scanner is automatically pre-processed to provide motion fields and myocardium segmentation. Via the web UI the results can then be analyzed and explored interactively.

Figure 2: Comparison of serial short-axis images acquired during exercise with a pedometer and rest. The function parameters are shown for the selected slice only for the user-selected heart cycles (white curve sections).

Figure 3: Blood pool volume of a patient with arrhythmia. The irregular cycles with reduced ejection fraction are clearly visible.

Figure 4: Blood velocity of a healthy volunteer during a Valsalva maneuver. The drop of the blood flow velocity can be seen in the interval from second 15 to 26.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)