Heart Rate related to Accuracy of left ventricular strain parameters derived from the single-shot compressed sensing cine CMR images of patients with cardiovascular disease

doi:10.21203/rs.3.rs-1734816/v1

Download PDF

Research Article

Heart Rate related to Accuracy of left ventricular strain parameters derived from the single-shot compressed sensing cine CMR images of patients with cardiovascular disease

https://doi.org/10.21203/rs.3.rs-1734816/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

Objective

To investigate the absolute and relative differences in the left ventricular (LV) global peak strain (GPS) values derived from single-shot compressed sensing (ss-CS) cine cardiac magnetic resonance (CMR) imaging and segmented cine CMR imaging (regarded as gold standard) of patients with cardiovascular disease, and investigate the potential patient-related factors that influence the misestimation of the GPS values from the ss-CS sequence.

Materials and Methods

Sixty-two cardiovascular disease patients with ss-CS cine and standard segmented cine CMR images were enrolled retrospectively. They were classified into different groups based on LV wall thickness and ejection fraction (EF). Absolute and relative differences in the LV GPS were calculated from both cine images. Stepwise multivariable linear regression analysis was performed to determine the patient-related factors that influenced the underestimation of the GPS values from ss-CS cine CMR images.

Results

LV GPS values from the ss-CS cine images were significantly underestimated. Absolute differences in the GPS values derived from both cine images varied significantly between the groups with different wall thickness and EFs (all P < 0.050). However, relative differences in the GPS values were not associated with the LV wall thickness or EF. The stepwise multivariable regression analysis showed that EF and heart rate were both associated with the absolute differences in the GPS values between the two cines (adjusted β-coefficient: 0.244 to 0.485, P < 0.05). Moreover, heart rate was independently associated with the relative differences in the global circumferential strain (GCS) and global longitudinal strain (GLS) values (GCS: β = 0.320, P = 0.011; GLS: β = 0.388, P = 0.002).

Conclusion

EF and heart rate were independently associated with the absolute differences in the LV GPS values. Heart rate was independently associated with the relative differences in the GPS values. Further improvements and standardization are necessary to increase the accuracy of the high-speed CMR imaging techniques in evaluating LV strain parameters.

Cardiovascular magnetic resonance

compressed sensing

left ventricular

strain

ejection fraction

heart rate

Conventional segmented cine cardiovascular magnetic resonance (CMR) imaging is the gold standard for evaluating left ventricular (LV) functions, including the strain parameters [1]. However, multiple breath-holds and regular heart rate (HR) is required to obtain high-quality images without artifacts from segmented cine CMR imaging [1]. This can be challenging in heart disease patients with arrhythmia and shortness of breath. Conventional real-time cine imaging with prospective ECG-gating and single-shot acquisition provides high-speed cine imaging with free-breathing and significantly eliminates breath motion artifacts[2]. However, reducing the duration of image acquisition decreases temporal and spatial resolution. Compressed sensing (CS) is an accelerated MRI technique that involves single-shot fast data acquisition with free-breathing and shows little or no reduction in the image quality compared with the conventional real-time cine imaging [3].

Single-shot CS (ss-CS) cine imaging significantly reduces examination time and enables accurate evaluation of the LV ejection fraction (EF) in patients with cardiac diseases when compared with the conventional segmented cine CMR imaging [4, 5]. However, LVEF is influenced by ventricular loading and does not directly reflect the status of myocardial contractility [6]. Myocardial strain detects the changes in LV function and subclinical myocardial dysfunction in a broad spectrum of cardiac conditions, and is a more sensitive marker of LV functional abnormalities [7]. Feature tracking (FT) is a novel technique to quantify myocardial strain from the segmented cine images [8]. CS provides cine images with reduced edge sharpness because of the characteristics of image acquisition [5]. Chen et al demonstrated that myocardial strain measurements from ss-CS cine images were significantly underestimated compared with those from the segmented cine images in volunteers [9]. The reasons for underestimating the global peak strains (GPSs) from the ss-CS cine images are unclear.

The ss-CS cine technique is commonly used to assess LV strain in various patients with different LV wall thickness (LVWT) and LV systolic functions [5, 10]. Therefore, it is necessary to evaluate the patient-related factors that influence the accuracy of LV strain parameters from ss-CS cine images of patients with cardiovascular diseases. Hence, in the present study, we compared LV myocardial strain parameters between segmented cine and ss-CS cine imaging. This study also analyzed the potential patient-related factors that are associated with the misestimation of LV myocardial strain using the ss-CS cine images.

Study subjects

Seventy-three patients that were diagnosed with thinned (LVWT<6mm), normal (9<LVWT<12mm), or thickened (LVWT>15mm) myocardium based on echocardiography [11-14] and underwent CMR at the First Affiliated Hospital of Nanjing Medical University between April 2020 and November 2020 were retrospectively enrolled in this study. Eleven patients were excluded from the study because of the following criteria: (1) common contraindications to MRI (n=0); (2) poor image quality due to arrhythmia and poor breath-holding (n=8), and (3) an interval of more than 1 week between echocardiography and CMR examinations (n=3). Finally, 62 patients were included in this study. We obtained written informed consent from all patients. This study was performed according to the protocol approved by the Ethics committee of the First Affiliated Hospital of Nanjing Medical University.

Cardiac magnetic resonance imaging protocols

All the patients underwent CMR imaging based on the segmented cine sequence (the standard of reference) and the ss-CS cine sequence in a clinical 3T MAGNETOM Skyra MRI scanner (Siemens Healthcare, Erlangen, Germany). The images were acquired along the short-axis (slice thickness of 8 mm with slice gap of 2 mm) and the long-axis views. The conventional segmented cine CMR used the retrospective electrocardiogram (ECG)-gated balanced steady-state free-precession (bSSFP) imaging sequence. The single-shot CS cine CMR used an adaptive prospective ECG-triggered sequence. The two cine sequences acquired images with 3.2 ms repetition time and 1.4 ms echo time. The temporal resolution was similar for both the sequences (segmented, 39 ms; ss-CS, 46 ms). The spatial resolution for both the sequences was 1.7*1.6 mm. End-inspiratory scanning was performed for all the subjects. The segmented cine imaging required 5-6 breath-holds and the ss-CS cine imaging required 1-2 breath-holds to cover the whole left ventricle. Adaptive prospective ECG-gating was performed during ss-CS cine imaging to record all the phases in a cardiac cycle [4, 5]. Twenty-five phases were designated for both the cine images so that post-processing was easier and uniform. The details of the imaging parameters for both the cine imaging sequences are shown in Supplementary Table 1.

CMR image analysis

All the CMR images were analyzed independently by two readers with at least two years of CMRI experience using the CVI 42 software (Circle Cardiovascular Imaging, Calgary, Canada).

Estimation of CMR image quality

The image quality was graded on a 4-point scale (1: nondiagnostic; 2: average; 3: good; 4: excellent) [10]. Images with clear endocardial and epicardial borders, clear contrast of the myocardium and the blood pool, clear papillary muscles, and valves, and without artifacts were graded as excellent. LV volumes and strains were evaluated for the images with scores ≥2.

Evaluation of LV volume and wall thickness

The LV volumetric analysis was performed by first tracing the LV endocardial and epicardial contours on a stack of short-axis images automatically. The contours were manually corrected, if necessary. The LV cavity included the papillary muscles and the endocardial trabeculations. The most basal short-axis slice included in the volumetry measurements was defined as the one with at least 270° of the LV circumference surrounded by the myocardium. The most apical short-axis slice included in the volumetric measurements was the last slice with a visible LV cavity. The left ventricular end-diastolic volume (EDV), end-systolic volume (ESV), stroke volume (SV), EF, and LV mass (LVM) were obtained from the segmented and ss-CS cine images for all the study subjects.

Estimation of LV strain

CMR feature tracking was used to measure the global peak strain (GPS) including global radial strain (GRS), global circumferential strain (GCS), and global longitudinal strain (GLS) for LV from both sequences. The contours of the LV endocardium and epicardium were drawn semi-automatically in the end-diastolic phase of the 2-, 3-, 4-chamber long axis and short-axis stacks and propagated automatically by the software throughout the cardiac cycle as previously described (11). Absolute and relative differences in the LV GPS values between the segmented cine and ss-CS cine images were calculated according to the following formulas: absolute difference (AD-GPS) = |GPScs – GPSsegmented|; relative difference (RD-GPS) = |GPScs – GPSsegmented| / |GPSsegmented|* 100%.

Statistical analysis

The data are represented as mean±standard deviation or median (interquartile range). Statistical analysis was performed using the following software: IBM SPSS Statistics for Windows version 23.0 (IBM Corp, Armonk, NY, USA); MedCalc Statistical software version 14.8 (MedCalc Software, Ostend, Belgium). Wilcoxon matched-pairs signed-rank test was used to compare the LV volume, LV strain, and image quality from the two cine images. The intraclass correlation coefficient (ICC) was used to compare the LV volume and strain parameters between the segmented and ss-CS cine CMR images. Bland-Altman analysis was used to evaluate the agreement in the LV GRS, GCS, and GLS values between the two cine imaging sequences. Kruskal-Wallis H test was used to compare the differences in the LV GPS values between the two cines (AD-GPSs and RD-GPSs) among the patient groups with different degrees of LVWT (thinned, normal, and thickened myocardium). Mann-Whitney U test was used to determine the differences in the LV GPS values from the cine images between patients with normal EF and reduced EF. Linear regression analysis was used to evaluate the correlation in the GRS values obtained from the segmented cine and ss-CS cine images in different LVWT groups. Stepwise multivariable linear regression analysis was performed to identify the patient-related factors that independently correlate with the underestimation of GPS values from the ss-CS cine images. The patient-related factors included sex, age, height, weight, body mass index, body surface area, HR, LVWT, and EF. ICC values were used to assess the inter-observer and intra-observer reproducibility. ICC values greater than 0.80, 0.60, and 0.40 were considered excellent, good, and moderate, respectively. P-value <0.05 was considered statistically significant.

Basic clinical characteristics of the study subjects

The sixty-two subjects (age, 50 ± 16 years) enrolled in this study included 32 males and 30 females. Their mean heights and weights were 166.26 ± 7.40 cm and 68.92 ± 14.80 Kg, respectively. Their mean body mass index, body surface areas, and heart rates were 24.81 ± 4.33 kg/m², 1.74 ± 0.22 m², and 71 ± 13 beats/min, respectively. The study subjects were diagnosed with dilated cardiomyopathy (n = 21), hypertrophic cardiomyopathy (n = 14), hypertensive cardiomyopathy (n = 6), valvular disease (n = 3), ventricular arrhythmia (n = 2), myocarditis (n = 3), secondary cardiomyopathy due to systemic diseases (n = 5), or apparent healthy heart (n = 8). The patients were classified into thinned myocardial wall (LVWT < 6mm; n = 22), normal myocardial wall (LVWT = 9 to 12mm; n = 19) and thickened myocardial wall (LVWT ≥ 15mm; n = 21) groups based on the LV wall thickness. The patients were divided into normal EF (EF ≥ 50%; n = 31) and reduced EF (EF < 50%; n = 31) groups based on the LV EF values derived from echocardiography (Fig. 1).

The acquisition time was 150 seconds for the segmented cine image and about 30 seconds for the ss-CS cine image. The image quality of the segmented cine images was significantly higher than the ss-CS images [segmented cine: 4 (3, 4) vs. ss-CS: 3(3, 4), P = 0.001] (Fig. 2). The intra-observer and inter-observer reproducibility for the LV volumetric and strain parameters was significantly high, with ICC value ≥ 0.953 for the segmented cine imaging and ICC value ≥ 0.834 for the ss-CS cine imaging (Supplementary Table 2).

LV GPS values from the ss-CS cine CMR images are significantly lower compared with those from the segmented cine CMR images

The LV volumetric parameters and GPS values between segmented and ss-CS cine images showed good or excellent correlation (ICC values ≥ 0.788). ESV and LVM values from the segmented cine and ss-CS cine images did not show significant differences (both P = 0.146). The EDV, SV, and EF values based on the ss-CS cine images were significantly lower than those from the segmented cine images, with median clinically negligible differences of 1.98 ml, 1.98 ml, and 0.84%, respectively (Table 1). Bland-Altman plots showed that the GRS, GCS, and GCS values from the ss-CS cine images were consistently and significantly lower compared with the segmented cine images (mean differences of -6.28%±5.53% in GRS, 2.01%±1.58% in GCS and 2.23%±2.38% in GLS respectively, Fig. 3).

Table 1

LV volumetric and strain parameters between segmented and ss-CS cine CMR imaging
Paramteres	Segmented cine	ss-CS cine	Segmented vs. ss-CS		P value	ICC
Paramteres	Segmented cine	ss-CS cine	Absolute Difference	Relative Difference(%)	P value	ICC
EDV (ml)	163.1 (129.9, 255.3)	161.4 (128.4, 250.7)	1.98 (0.76, 5.27)	0.97 (0.42, 3.14)	< 0.001	0.999
ESV (ml)	77.1 (43.7, 192.6)	75.9 (43.1, 192.5)	0.74 (0.34, 1.73)	0.79 (0.39, 2.04)	0.146	0.999
SV (ml)	72.4 (61.9, 93.5)	72.6 (59.8, 92.5)	1.98 (0.86, 4.95)	2.83 (1.01, 5.87)	0.002	0.982
EF (%)	46.5 (26.9, 62.6)	45. 6 (27.2, 62.6)	0.84 (0.28, 1.89)	1.92 (0.66, 4.21)	0.015	0.996
LVM (mg)	132.8 (97.8, 176.8)	133.9 (98.8, 177.1)	6.54 (2.32,10.80)	4.46 (1.59, 8.22)	0.146	0.987
GRS (%)	19.1 (10.5, 27.3)	11.8 (7.6, 20.9)	5.25 (2.83, 7.78)	26.55 (16.60, 40.40)	< 0.001	0.862
GCS (%)	-12.6 (-19.4, -8.3)	-11.3 (-16.3, -7.2)	1.75 (0.93, 3.14)	14.94 (7.43, 20.12)	< 0.001	0.960
GLS (%)	-8.5 (-11.9, -5.7)	-6.2 (-8.6, -4.3)	1.86 (1.04, 3.87)	28.20 (15.26, 38.26)	< 0.001	0.788
The data are represented as the median (first quartile, third quartile). Wilcoxon matched-pairs signed-rank test was used to compare LV volume and strain between segmented and ss-CS cine CMR. ICC was used to evaluate the correlation of all LV parameters between the two cine images. LV = left ventricular, ss-CS = single shot compressed sensing, CMR = cardiovascular magnetic resonance, EDV = end-diastolic volume, ESV = end-systolic volume, SV = stroke volume, EF = ejection fraction, LVM = left ventricular mass, GRS = global radial strain, GCS = global circumferential strain, GLS = global longitudinal strain, ICC = intraclass correlation coefficient

Significant Influence of Wall thickness and LV function on AD-GPSs but not on RD-GPS between the two cine images

The absolute differences in the GRS and GCS values between segmented cine and ss-CS cine images were significantly higher for patients with thickened myocardium compared with those with thinned myocardium (AD-GRS, 5.80% vs 3.12%, P = 0.047; AD-GCS, 2.42% vs 1.36%, P = 0.029; Table 2). However, the relative differences in the GPS values (RD-GRS, RD-GCS, and RD-GLS) between segmented cine and ss-CS cine images did not show significant differences among patients with normal, thinned, or thickened myocardium (Table 2). Moreover, the absolute differences in the GPS values (AD-GRS, AD-GCS, and AD-GLS) were significantly higher between segmented cine and ss-CS cine images for the patients with normal EF compared with those with lower EF, but the relative differences in the GPS values (RD-GRS, RD-GCS, and RD-GLS) between segmented cine and ss-CS cine images were similar for patients with normal and lower EF (Table 2). The correlation between segmented cine-derived GRS and ss-CS cine-derived GRS was lower for the thinned myocardium compared with the normal myocardium (R_thinned vs. R_normal: 0.548 vs. 0.888, P = 0.019) and the thickened myocardium (R_thinned vs. R_thickened: 0.548 vs.0.891, P = 0.014) (Fig. 4).

Table 2

Absolute and relative differences in the LV GRS, GCS, and GLS values between the segmented and ss-CS cine images of patients with distinct types of wall thickness and ejection fractions.
		Normal myocardium (n = 19)	Thinned myocardium (n = 22)	Thickened myocardium (n = 21)	P value	EF ≥ 50% (n = 31)	EF < 50% (n = 31)	P value
absolute difference	GRS	5.57 (3.41, 7.84)	3.12 (0.88, 6.40)	5.80 (3.33, 11.35)	0.033	6.30 (4.55, 11.86)	3.41 (1.46, 5.57)	0.001
	GCS	1.97 (0.54, 4.06)	1.36 (0.42, 1.95)	2.42 (1.12, 3.65)	0.031	2.63 (1.59, 4.07)	1.13 (0.38, 1.94)	< 0.001
	GLS	2.82 (1.02, 4.89)	1.38 (0.89, 2.67)	2.26 (1.25, 3.79)	0.198	2.82 (1.10, 4.47)	1.33 (0.58, 2.45)	0.021
relative difference	GRS	21.21 (15.45, 36.41)	26.50 (12.10, 47.25)	29.02 (21.11, 47.60)	0.240	26.56 (19.18, 40.14)	26.24 (16.39, 46.65)	0.849
	GCS	12.70 (4.95, 18.86)	14.33 (8.94, 22.48)	17.22 (8.62, 23.73)	0.172	15.35 (7.52, 19.32)	14.21 (4.52, 22.44)	0.578
	GLS	26.91 (10.53, 36.64)	26.90(18.06, 38.94)	32.68 (15.92, 43.24)	0.584	27.82(10.06, 38.18)	26.52 (7.88, 35.53)	0.522
The data are represented as the median (first quartile, third quartile). Kruskal-Wallis H test was used to compare the differences in the LV strains between the two cine images (absolute and relative differences) of patients with different LVWTs. Mann-Whitney U test was used to compare the differences in the LV strains between the two cine images of patients from different EF groups. LV = left ventricular, ss-CS = single shot compressed sensing, GRS = global radial strain, GCS = global circumferential strain, GLS = global longitudinal strain, EF = ejection fraction, LVWT = left ventricular wall thickness

EF and heart rate are associated with the underestimation of GPS from ss-CS cine CMR images

Stepwise multivariable linear regression analysis was performed to determine the patient-related factors (sex, age, height, weight, body mass index, body surface area, HR, LVWT, and EF) that contribute to the underestimation of GPS values derived from the ss-CS cine CMR images. EF and HR were independently associated with the absolute differences in the GPS values between the segmented cine and ss-CS cine images (Table 3). Furthermore, HR was associated with the relative differences in the GCS and GLS values between the segmented cine and ss-CS sine images (RD-GCS: β = 0.320, P = 0.011; RD-GLS: β = 0.388, P = 0.002) (Fig. 5). The predicted GPSsegmented value for patients with only ss-CS cine images is calculated with the following formula: GPSsegmented = GPScs / (RD-GPS% +1), where RD-GCS = 0.240*HR − 1.623, RD-GLS = 0.436*HR − 2.856, and RD-GRS = 26.55. This formula requires only GPScs and HR to calculate the GPSsegmented values.

Table 3

Patient-related factors independently associated with the absolute and relative differences in LV global peak strains between segmented and ss-CS cine (only significant independent predictors in stepwise multivariable linear regression analysis were listed)
Independent variables	Factors in models		B	SD	β	P Value	95% confidence interval
Independent variables	Factors in models		B	SD	β	P Value	Lower limit	Upper limit
AD-GRS		EF	4.259	1.289	0.392	0.002	1.680	6.839
AD-GCS		EF	1.437	0.323	0.485	< 0.001	0.790	2.084
		HR	0.033	0.012	0.292	0.010	0.008	0.058
AD-GLS		EF	1.415	0.464	0.361	0.003	0.487	2.343
		HR	0.037	0.018	0.244	0.044	0.001	0.072
RD-GRS		None
RD-GCS		HR	0.240	0.092	0.320	0.011	0.056	0.423
RD-GLS		HR	0.436	0.134	0.388	0.002	0.168	0.704
Patient-related factors including sex, age, height, weight, body mass index, body surface area, HR, LVWT, and EF were evaluated in each stepwise multivariable linear regression analysis. LV = left ventricular, ss-CS = single shot compressed sensing, AD = absolute difference, RD = relative difference, GRS = global radial strain, GCS = global circumferential strain, GLS = global longitudinal strain, EF = ejection fraction, HR = heart rate, LVWT = left ventricular wall thickness

This study demonstrated that the LV GPS values derived from the ss-CS cine images were significantly lower than those derived from the segmented cine images. Furthermore, GPS values for the thinned myocardium derived from the two cine images showed poor correlation compared with the normal and thickened myocardium. Higher EF and HR values independently augmented the absolute differences in the GPS values between the segmented and ss-CS cine images. In addition, relative differences in the GCS and GLS values were augmented by increasing HR. GPS values for segmented cine can be calculated for patients with only ss-CS cine images using the HR and GPScs values.

Previous studies have shown that ss-CS cine CMR image quality is high and LV volumetric assessments from the ss-CS cine images with breath-holding or free-breathing are comparable to those from segmented cine images [5, 10, 15]. Our study showed strong agreement and high reproducibility of the LV volumetric parameters between the two cine imaging sequences. This confirmed that ss-CS cine imaging is a dependable clinical method to evaluate LV volumetric parameters in patients with different cardiac functional states and LVWTs.

Chen et al demonstrated that LV GRS, GCS, and GLS derived from the ss-CS cine images of healthy volunteers were significantly lower compared with those derived from the segmented cine images [9]. In this study, although the LV GPS values from the segmented cine and ss-CS cine images were lower for the patients with cardiac dysfunction compared with the healthy volunteers, the relative differences in the LV GPS values derived from the segmented cine and ss-CS cine images were similar to the report by Chen et al (GRS: 26.55%, GCS: 14.46%, and GLS: 27.10%). The relative differences in the GCS values from the two cine images were lower than the differences in the GRS and GLS values. This observation was also consistent with the study by Chen et al [9]. In contrast to our findings, Kido et al did not observe any significant differences in the LV GRS values derived from the ss-CS cine images using 80 iterations and conventional cine images [16]. We used 40 iterations in this study. Temporal resolution is important for analyzing the strain parameters. In our study, the differences in the temporal resolution between segmented cine imaging (39 ms) and ss-CS cine imaging (46 ms) were not significant [17]. Therefore, the differences in the number of iterations may have resulted in the contradictory conclusions between our study and the study by Kido et al. This suggested that the ss-CS cine technique, including the number of iterations, should be standardized for clinical practice in the future.

Stokke et al demonstrated that LVEF was highly dependent on the longitudinal and circumferential shortening, thereby suggesting a positive correlation between global strain and EF [18]. In our study, patients with higher EF showed higher absolute differences but similar relative differences in the LV GPS values derived from segmented cine and ss-CS cine CMR images. This demonstrated that the GPS values from the ss-CS cine CMR images and the segmented cine CMR images were proportional for all patients irrespective of their EF status.

Cho et al reported different GLS values for severe aortic stenosis patients with different wall thicknesses, thereby suggesting a relationship between wall thickness and myocardial strain [19]. Rodrigues et al demonstrated that LV wall thickening was associated with reduced myocardial longitudinal and circumferential shortening as measured by tagging CMR imaging in patients with hypertensive heart disease [20]. Our study demonstrated that the absolute differences in the GRS and GCS values varied according to different degrees of wall thickness. Furthermore, the correlation of GRS values from the CMR images of patients with thinned myocardium was lower compared with the correlation of GRS values from CMR images of patients with normal and thickened myocardium. This suggested that erroneous definition of the endocardial and epicardial borders significantly varied the GRS values compared with the GCS and GLS values [21].

Boettler et al. showed that HR significantly influenced the GPS values in children (21). GPS values showed a negative correlation with HR; that is, higher GPS values correlated with decreasing HR [22]. Our findings showed that the HR was significantly associated with the absolute and relative differences in the GPS values between the two CMR cine sequences. Our data were consistent with previously reported findings with the real-time cine imaging techniques [2, 9]. This suggested that the raw data acquired by the ss-CS cine imaging sequence was underestimated when acquired at a higher HR. Higher heart rates significantly shortened the cardiac phase and reduced image quality [23]. Lower image quality resulted in the underestimation of the strain values [24].

Our study demonstrated that the absolute differences in the GPS values between the two cine imaging sequences were influenced by EF and HR. However, relative differences in the GPS values between the two cine imaging sequences was influenced only by HR. We did not observe any relationship between other patient-related clinical factors such as sex, age, height, weight, body mass index, body surface area, LVWT, and EF and the relative differences in the GPS values from the two cine imaging sequences. This data was consistent with the previous CS cine findings in healthy volunteers and other real-time cine CMR imaging techniques [2, 9]. This suggested that relative adjustments during LV strain analysis of the ss-CS cine images may be clinically relevant.

Our study has several limitations. First, the quality of the ss-CS cine images varied with the number of iteration reconstructions [25]. In this study, we performed 40 iterations. The relationship between the number of iterations and the accuracy of LV strain measurements should be investigated in future studies. Second, an interpolation is necessary during post-processing to ensure the same number of phases. However, this causes lower edge sharpness, which may result in the underestimation of LV GPS values from the ss-CS cine CMR images. Moreover, underestimation of the myocardial strain by ss-CS cine CMR is highly dependent on the CS reconstruction methods and acquisition parameters. Third, patient-related factors that influence LV strain values may vary according to the acquisition parameters and reconstruction methods used. Therefore, refinement of CS image acquisition and reconstruction techniques may require further analysis of the patient-related factors and may not be the same as reported by our study. The patient-related factors influencing the underestimation of GPS values are applicable to the ss-CS technique used in our study.

This study demonstrated that ss-CS cine CMR imaging significantly underestimated the LV strain parameters in cardiovascular patients compared with the traditional segmented cine CMR imaging. EF and HR were independently associated with the underestimation of LV strain parameters by ss-CS cine imaging. Furthermore, we derived a formula to estimate segmented GPS in patients with only ss-CS cine CMR data using only GPScs and HR values. Therefore, further improvements and standardization is required to accurately calculate LV strain parameters from high-speed CMR imaging techniques.

LV: left ventricular

CS: compressed sensing

CMR: cardiovascular magnetic resonance

EF: ejection fraction

GPS: global peak strain

GRS: global radial strain

GCS: global circumferential strain

GLS: global longitudinal strain

FT: feature tracking

WT: wall thickness

HR: heart rate

EDV: end-diastolic volume

ESV: end-systolic volume

SV: stroke volume

LVM: left ventricular mass

AD: absolute difference

RD: relative difference

ICC: intraclass correlation coefficient

Funding

This research was partly supported by the National Natural Science Foundation under grants (81701651 and 81601464).

Acknowledgements

The authors of this manuscript declare no impact on results from funding supporters.

Competing interests

The authors have no relevant financial or non-financial interests to disclose.

Author Contributions

Yinsu Zhu, Yi Xu and Xiaomei Zhu contributed to the study conception and design. Material preparation, data collection and analysis were performed by Yang Chen, Panpan Xu, Jun Wang, Wen Qian and Wangyan Liu. The first draft of the manuscript was written by Yang Chen and Panpan Xu and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Ethics approval

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of Nanjing Medical University.

Consent to participate

Informed consent was obtained from all individual participants included in the study.

Consent to publish

The authors affirm that human research participants provided informed consent for publication of the images in Figure2.

Axel L, Otazo R (2016) Accelerated MRI for the assessment of cardiac function. Br J Radiol 89:20150655. http://doi.org/10.1259/bjr.20150655.
Zhu X, Schwab F, Marcus R, Hetterich H, Theisen D, Kramer H, et al (2015) Feasibility of free-breathing, GRAPPA-based, real-time cardiac cine assessment of left-ventricular function in cardiovascular patients at 3T. Eur J Radiol 84:849-855. http://doi.org/10.1016/j.ejrad.2015.02.016.
Delattre BMA, Boudabbous S, Hansen C, Neroladaki A, Hachulla AL, Vargas MI (2020) Compressed sensing MRI of different organs: ready for clinical daily practice? Eur Radiol 30:308-319. http://doi.org/10.1007/s00330-019-06319-0.
Kido T, Kido T, Nakamura M, Watanabe K, Schmidt M, Forman C, et al (2016) Compressed sensing real-time cine cardiovascular magnetic resonance: accurate assessment of left ventricular function in a single-breath-hold. J Cardiovasc Magn Reson 18:50. http://doi.org/10.1186/s12968-016-0271-0.
Vermersch M, Longere B, Coisne A, Schmidt M, Forman C, Monnet A, et al (2020) Compressed sensing real-time cine imaging for assessment of ventricular function, volumes and mass in clinical practice. Eur Radiol 30:609-619. http://doi.org/10.1007/s00330-019-06341-2.
Konstam MA, Abboud FM (2017) Ejection Fraction: Misunderstood and Overrated (Changing the Paradigm in Categorizing Heart Failure). Circulation 135:717-719. http://doi.org/10.1161/CIRCULATIONAHA.116.025795.
Pryds K, Larsen AH, Hansen MS, Grondal AYK, Tougaard RS, Hansson NH, et al (2019) Myocardial strain assessed by feature tracking cardiac magnetic resonance in patients with a variety of cardiovascular diseases - A comparison with echocardiography. Sci Rep 9:11296. http://doi.org/10.1038/s41598-019-47775-4.
Ananthapadmanabhan S, Vo G, Nguyen T, Dimitri H, Otton J (2021) Direct comparison of multilayer left ventricular global longitudinal strain using CMR feature tracking and speckle tracking echocardiography. BMC Cardiovasc Disord 21:107. http://doi.org/10.1186/s12872-021-01916-8.
Chen Y, Qian W, Liu W, Zhu Y, Zhou X, Xu Y, et al (2021) Feasibility of single-shot compressed sensing cine imaging for analysis of left ventricular function and strain in cardiac MRI. Clin Radiol 76:471.e471-471.e477. http://doi.org/doi: 10.1016/j.crad.2020.12.024.
Wang J, Lin Q, Pan Y, An J, Ge Y (2020) The accuracy of compressed sensing cardiovascular magnetic resonance imaging in heart failure classifications. Int J Cardiovasc Imaging 36:1157-1166. http://doi.org/10.1007/s10554-020-01810-y.
Shub C, Klein AL, Zachariah PK, Bailey KR, Tajik AJ (1994) Determination of Left Ventricular Mass by Echocardiography in a Normal Population: Effect of Age and Sex in Addition to Body Size. Mayo Clinic Proceedings 69:205-211. http://doi.org/10.1016/s0025-6196(12)61058-1.
Lee PT, Dweck MR, Prasher S, Shah A, Humphries SE, Pennell DJ, et al (2013) Left ventricular wall thickness and the presence of asymmetric hypertrophy in healthy young army recruits: data from the LARGE heart study. Circ Cardiovasc Imaging 6:262-267. http://doi.org/10.1161/CIRCIMAGING.112.979294.
Galand V, Ghoshhajra B, Szymonifka J, Das S, Orencole M, Barre V, et al (2020) Left ventricular wall thickness assessed by cardiac computed tomography and cardiac resynchronization therapy outcomes. Europace 22:401-411. http://doi.org/10.1093/europace/euz322.
Sjogren AL (1971) Left ventricular wall thickness determined by ultrasound in 100 subjects without heart disease. Chest 60:341-346. http://doi.org/10.1378/chest.60.4.341.
Kido T, Kido T, Nakamura M, Watanabe K, Schmidt M, Forman C, et al (2017) Assessment of Left Ventricular Function and Mass on Free-Breathing Compressed Sensing Real-Time Cine Imaging. Circ J 81:1463-1468. http://doi.org/10.1253/circj.CJ-17-0123.
Kido T, Hirai K, Ogawa R, Tanabe Y, Nakamura M, Kawaguchi N, et al (2021) Comparison between conventional and compressed sensing cine cardiovascular magnetic resonance for feature tracking global circumferential strain assessment. J Cardiovasc Magn Reson 23:10. http://doi.org/10.1186/s12968-021-00708-5.
Backhaus SJ, Metschies G, Billing M, Schmidt-Rimpler J, Kowallick JT, Gertz RJ, et al (2021) Defining the optimal temporal and spatial resolution for cardiovascular magnetic resonance imaging feature tracking. J Cardiovasc Magn Reson 23:60. http://doi.org/10.1186/s12968-021-00740-5.
Stokke TM, Hasselberg NE, Smedsrud MK, Sarvari SI, Haugaa KH, Smiseth OA, et al (2017) Geometry as a Confounder When Assessing Ventricular Systolic Function: Comparison Between Ejection Fraction and Strain. J Am Coll Cardiol 70:942-954. http://doi.org/10.1016/j.jacc.2017.06.046.
Cho EJ, Park SJ, Kim EK, Lee GY, Chang SA, Choi JO, et al (2017) Effects of increased left ventricular wall thickness on the myocardium in severe aortic stenosis with normal left ventricular ejection fraction: Two- and three-dimensional multilayer speckle tracking echocardiography. Echocardiography 34:511-522. http://doi.org/10.1111/echo.13474.
Rodrigues JC, Rohan S, Dastidar AG, Trickey A, Szantho G, Ratcliffe LE, et al (2016) The Relationship Between Left Ventricular Wall Thickness, Myocardial Shortening, and Ejection Fraction in Hypertensive Heart Disease: Insights From Cardiac Magnetic Resonance Imaging. J Clin Hypertens (Greenwich) 18:1119-1127. http://doi.org/10.1111/jch.12849.
Rosner A, Barbosa D, Aarsaether E, Kjonas D, Schirmer H, D'Hooge J (2015) The influence of frame rate on two-dimensional speckle-tracking strain measurements: a study on silico-simulated models and images recorded in patients. Eur Heart J Cardiovasc Imaging 16:1137-1147. http://doi.org/10.1093/ehjci/jev058.
Boettler P, Hartmann M, Watzl K, Maroula E, Schulte-Moenting J, Knirsch W, et al (2005) Heart rate effects on strain and strain rate in healthy children. J Am Soc Echocardiogr 18:1121-1130. http://doi.org/10.1016/j.echo.2005.08.014.
Boll DT, Gilkeson RC, Merkle EM, Fleiter TR, Duerk JL, Lewin JS (2005) Functional cardiac CT and MR: effects of heart rate and software applications on measurement validity. J Thorac Imaging 20:10-16. http://doi.org/10.1097/01.rti.0000154077.13947.8e.
Shiran A, Blondheim DS, Shimoni S, Jabarren M, Rosenmann D, Sagie A, et al (2019) Effect of image quality on accuracy of two-dimensional strain echocardiography for diagnosing ischemic chest pain: a 2DSPER multicenter trial substudy. Int J Cardiovasc Imaging 35:617-625. http://doi.org/10.1007/s10554-018-1495-x.
Goebel J, Nensa F, Bomas B, Schemuth HP, Maderwald S, Gratz M, et al (2016) Real-time SPARSE-SENSE cardiac cine MR imaging: optimization of image reconstruction and sequence validation. Eur Radiol 26:4482-4489. http://doi.org/10.1007/s00330-016-4301-y.

No competing interests reported.

SupplementaryTables.docx

Download PDF

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

Heart Rate related to Accuracy of left ventricular strain parameters derived from the single-shot compressed sensing cine CMR images of patients with cardiovascular disease

Status:

Version 1

Abstract

Objective

Materials and Methods

Results

Conclusion

Figures

Introduction

Materials And Methods

Results

Discussion

Conclusions

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1