Temporal interpolation of real-time cine images for ventricular function assessment
Journal of Cardiovascular Magnetic Resonance volume 16, Article number: O42 (2014)
In uncooperative patients or in presence of arrhythmia, real-time free-breathing cine MR approaches may offer an alternative to well-established segmented data acquisition strategies [1, 2]. Segmented acquisition strategies collect data throughout a number of heartbeats and typically a fixed number of cardiac phases are reconstructed for each slice representing an average RR interval. In real-time cine imaging, the number of reconstructed phases may be different for each slice due to heart rate variations. However, it is desirable to reconstruct a predefined number of cardiac phases per slice to facilitate functional analysis and processing. In this work, we present an image reconstruction approach that retrospectively interpolates real-time cine images to calculate a predefined number of cardiac phases per heartbeat and slice. Ejection fraction is compared between segmented cine images and interpolated real-time cine images in healthy volunteers.
Segmented breath-hold and real-time free-breathing cine images were acquired in four healthy volunteers on a clinical 3T MR scanner (MAGNETOM Skyra, Siemens Healthcare, Erlangen, Germany). Acquisition parameters for each volunteer are represented in Table 1. FOVs for real-time studies were kept as small as possible to improve spatial resolution. T-PAT factor 3 was used to improve temporal resolution. The linear interpolation module was fully integrated into the reconstruction pipeline of the scanner. During imaging, heart rate was continuously monitored and real-time images were linearly interpolated to a fixed number of cardiac phases for each heartbeat and for each slice (Figure 1). 20 cardiac phases were calculated per heartbeat by interpolation during real-time acquisitions to achieve 50 ms nominal temporal resolution. Subsequently, interpolated real-time images were analyzed to evaluate LV function (syngo Argus 4D ventricular function, Siemens Healthcare, Erlangen, Germany) and compared with results based on segmented images.
Interpolated real-time images were loaded to Argus and endo-cardial borders were detected prior to analysis. EF measurements (see Table 2) from interpolated real-time and segmented images were found to be in good agreement. End-systolic and end-diastolic phases for segmented, non-interpolated real-time, and interpolated real-time images are given in Figure 2. Interpolation increased the SNR, but introduced slight blurriness. The level of blurring was low and thus it did not affect the functional evaluation.
In this feasibility study we presented initial evidence that interpolated real-time cine images may be used to facilitate cardiac functional analysis using well-established post-processing software. The current implemented algorithm will be extended to take arrhythmia events into account.
The author is a full-time employee of Siemens Healthcare USA, Inc.
Carr JC, et al: Radiology. 2001, 219: 828-834. 10.1148/radiology.219.3.r01jn44828.
Hori Y, et al: J Card Magn Reson. 2003, 5: 439-450. 10.1081/JCMR-120022260.
About this article
Cite this article
Saybasili, H., McNeal, G.R., Zuehlsdorff, S. et al. Temporal interpolation of real-time cine images for ventricular function assessment. J Cardiovasc Magn Reson 16 (Suppl 1), O42 (2014). https://doi.org/10.1186/1532-429X-16-S1-O42