Introduction
Myocardial strain analysis plays a key role in the quantitative assessment of global left ventricular (LV) function in various cardiac pathologies and provides prognostic information over conventional parameters [
1]. Heart failure (HF) patients with left bundle branch block (LBBB) show regional timing differences causing an imbalanced contractile function throughout the LV and, in particular, across the septal and lateral wall [
2‐
4]. Segmental strains can be used to calculate parameters of mechanical discoordination (opposing shortening and stretching within the LV). This paradoxical wall deformation underlies an inefficient pump function and is considered to be the functional substrate amenable to resynchronization. Therefore, segmental strain measurements are increasingly recognized for their additional value in the selection of HF patients for cardiac resynchronization therapy (CRT) [
4‐
7].
Echocardiographic techniques, such as the speckle tracking method, provide strain measures on a global and segmental scale. Although echocardiography is widely available, strain analysis is highly dependent on the image quality provided by the acoustic window and remains limited because of its reproducibility [
8,
9]. In the meantime, cardiovascular magnetic resonance (CMR) has rapidly emerged as a robust imaging modality with high accuracy to provide detailed information on cardiac morphology, function and tissue characterisation [
10]. In particular, CMR myocardial tissue tagging (CMR-TAG) produces high quality strain measures on a global and segmental level. It is considered to be the gold standard, but is not widely available and is predominantly used for scientific purposes [
11]. On the other hand, CMR cine imaging is part of every standard clinical protocol and is increasingly utilized in the screening of CRT candidates to determine the LV ejection fraction. Therefore, CMR feature-tracking (CMR-FT) post-processing techniques have been developed to derive strain measurements from standard cine images [
12]. Although global strain measures show reasonable agreement with the CMR-TAG technique, segmental strain measures have repeatedly been proven to be insufficient [
13‐
17]. This study evaluates the novel segment length in cine (SLICE) post-processing technique on standard CMR cine images to derive accurate segmental strain measures in CRT candidates without the use of commercial software.
Discussion
This study demonstrates that the SLICE post-processing technique on standard CMR cine images offers both accurate and robust circumferential strain measures compared to the gold standard CMR-TAG technique in CRT candidates. Strain measures on a segmental scale, particularly of interest in this patient group to predict CRT response, show close agreement with the CMR-TAG technique, whereas recently published data on CMR-FT software showed disappointing results [
13‐
17].
Chronic HF patients with LBBB, eligible for CRT, show large diversity in the manifestation of mechanical timing differences (dyssynchrony) and deformation abnormalities (discoordination) throughout the LV. Traditionally, echocardiographic modalities offer high temporal resolution imaging, which enables the assessment of regional timing differences throughout the LV. However, echocardiographic dyssynchrony parameters showed disappointing results in the PROSPECT study [
8]. More recent, single-centre studies showed mechanical discoordination rather than mechanical dyssynchrony to be predictive for CRT response [
4‐
7]. In this study, quantification of mechanical discoordination showed wide variation among CRT candidates, ranging from a normally coordinated to a severely discoordinated LV contraction pattern, with predominant stretching of the septum during systole (LBBB-2 pattern). Because of this wide interperson variation in segmental strains and discoordination rates, segmental strain analysis holds the potential to discriminate between CRT responders and non-responders at baseline. Using SLICE, the presence of a paradoxical septum movement was distinctly noticeable by qualitative assessment of the SLICE strain-curve (Fig.
1a). Classification of septum strains to pre-specified strain patterns (Fig.
3) showed good agreement between the SLICE and CMR-TAG method. Classification of a pseudonormal (LBBB-3) septum strain pattern is of clinical relevance as this pattern is associated with smaller benefit of CRT compared to other septum strain patterns [
25]. Quantification of strain patterns by measuring regional systolic strains, the cumulative amount of SRS, SSI and the discoordination rate ISF showed good to excellent agreement with the CMR-TAG technique (Figs.
4 and
5). These are conventional markers of mechanical discoordination that have been shown to be accurate in the prediction of CRT response in multiple single centre studies [
4‐
6,
23]. In addition, sensitivity of the SLICE technique proved to be high enough to detect the differences in discoordination values that were found between IMCP patients and NICMP patients. Patients with an ICMP have significantly lower discoordination values compared to NICMP patients, which might explain the lack of CRT response in this subgroup of patients. In general, the SLICE technique tends to produce slightly lower discoordination values compared to the CMR-TAG standard. This might be due to the lower temporal resolution that was used for cine imaging compared to CMR-TAG (~50 ms vs. ~15 ms) being less sensitive for peak strain values [
28]. It is possible that this shortcoming in temporal resolution had a larger effect on the assessment of time-based dyssynchrony parameters, which could explain the lack of agreement and reproducibility that was found for these parameters. However, dyssynchrony parameters have limited predictive value for the selection of CRT patients [
8]. Therefore, future studies should focus on the classification of septal strain patterns and discoordination markers, which can be accurately derived by the SLICE post-processing technique. Also, accelerated CMR imaging techniques such as parallel imaging can overcome this problem by improving the spatiotemporal resolution of the cine image acquisition [
29]. However, since the SLICE technique is intended as a post-processing technique on standard cine images, a temporal resolution of ~50 ms was used for the acquisition of cine images as this is typical in most standard clinical protocols.
Conceptual differences between the SLICE and CMR-TAG techniques should be considered. CMR-TAG sequences cover the LV with magnetization saturated bands in a grid format at end-diastole [
30]. Post-processing of these ‘tagged’ images by tracing the displacement of the taglines throughout the cardiac cycle produces high quality measures of intramural shortening. On the other hand, SLICE is performed by measuring total segment length between two anatomical landmarks throughout all phases on standard cine images. This method produces a measure of relative frame-to-frame segment length change that, in essence, approximates the net result of all intramural strains combined throughout the segment. A disadvantage of the SLICE method is that it derives strain measures from the apparent in-plane motion of the anatomical landmarks. However, this apparent in-plane movement may also be caused by through-plane displacements of oblique or tapering structures that form the anatomical landmarks. For this reason we could only analyse the mid-LV slice, since this plane is relatively motion independent. CMR-FT software neglects these pitfalls and the user is not able to track and trace the analysis steps. This might be the reason for the previous failure of segmental strain analysis by the CMR-FT software. Another consequential disadvantage is the inability to determine LV torsion, as this analysis requires both basal and apical rotation measurements.
To our knowledge, this is the first study to demonstrate a post-processing technique on standard CMR cine images that offers both accurate and robust circumferential strain measures on a segmental scale in CRT candidates. Both the visual classification of septal strain patterns and the quantification of conventional discoordination parameters being SRS, SSI and ISF showed close agreement with the ‘gold standard’ CMR-TAG technique, bringing the use of accurate predictors of CRT response a step closer to clinical practice.
Some limitations need to be addressed. Firstly, the present study was specified to HF patients with LBBB only. However, this might also be considered a strength of this study since segmental strains are particularly of interest in this specific population. Secondly, some geographic selection bias may have occurred as only study participants that were included close to our centre were invited to participate in the present CMR sub-study. Furthermore, the processing time for the CMR-TAG analysis was less than 20 min compared to a maximum of 60 min for the SLICE method. Although the SLICE technique was more time-consuming in the present study, previously used CMR-TAG post-processing software techniques also proved to be time-consuming and less user-friendly compared to the new SinMod technique by inTag.
For future studies, the SLICE processing time can be substantially reduced (halved) by limiting the analysis to the systolic phase, since diastolic strains are not incorporated in discoordination markers (Fig.
2). In fact, segmental systolic strains can already be calculated after analysing two frames providing both end-diastolic and end-systolic segment length (Fig.
1a). The number of frames that are required per strain parameter is displayed in Fig.
S2 in the Supplemental Material. Further reduction of processing time, as well as improvement of reproducibility and clinical applicability, can be achieved by automated assessment. Additionally, implementing radial taglines to standard cine imaging will facilitate the detection of the landmarks, thus further enhancing strain analysis. Ultimately, SLICE might serve as a new principle for integration in CMR-FT algorithms.
In conclusion, the novel SLICE post-processing technique requires standard cine images only and provides both accurate and robust myocardial strain measures on a segmental scale in HF patients with LBBB. Future studies will focus on the prognostic value of these strain measures in CRT candidates.