Biventricular dysfunction and blood oxygenation deficits in obstructive sleep apnea: a prospective study with non-contrast cardiac MRI

Abstract

Background

Conventional cardiac magnetic resonance imaging(CMR) metrics may remain normal in Obstructive Sleep Apnea (OSA) despite subclinical myocardial injury, limiting early risk stratification. This study aimed to evaluate whether biventricular function, mechanics, and tissue characteristics assessed by non-contrast CMR differ between patients with severe and non-severe OSA, and to explore their association with markers of nocturnal hypoxemia.

Methods

Seventy-five newly diagnosed OSA patients(62 male; age 43.7 ± 9.3 years) were prospectively included and underwent polysomnography followed by non-contrast CMR, including cine imaging, native T1 and T2 mapping. Patients were stratified by apnea-hypopnea index (AHI) into non-severe (n = 21, 4 mild and 17 moderate cases) and severe (n = 54) groups. Biventricular parameters were compared and correlated with AHI and the oxygen desaturation index (ODI).

Results

Compared with non-severe OSA group, severe OSA group exhibited higher left ventricular mass (LVM) and lower left ventricular global circumferential and radial strain (all p < 0.05). Right ventricular (RV) dysfunction was more pronounced, with significantly lower in global longitudinal, circumferential, and radial strain and strain rates (all p < 0.05). The RV end-systolic remodeling index (RVESRI) was higher in severe OSA (p = 0.002). The right ventricular blood pool T2 value (RVT2) was significantly lower in severe OSA than in the non-severe OSA group and moderately correlated with AHI (ρ=-0.435) and ODI (ρ=-0.425). Collectively, multiple CMR parameters showed weak‑to‑moderate correlations with AHI and ODI (ρ ranging from − 0.301 to 0.476), indicating that OSA severity is associated with a broad spectrum of subclinical biventricular alterations rather than a single dominant abnormality.

Conclusion

Severe OSA is associated with subclinical biventricular dysfunction, disproportionately affecting the RV. Elevated RVESRI may indicate early systolic maladaptation, while lower RVT2 shows promise as a non-invasive imaging marker associated with hypoxemic burden, warranting further investigation. Non-contrast CMR enables detection of cardiac injury, offering valuable potential for risk stratification in OSA patients.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12880-026-02478-x.

Introduction

Obstructive sleep apnea (OSA) is a prevalent sleep-related breathing disorder characterized by repetitive upper airway collapse during sleep, leading to intermittent hypoxia and sleep fragmentation. In a multi-ethnic cohort, the prevalence among Chinese reached 66.4% [1]. Growing evidence indicates that OSA exerts profound effects on the cardiovascular system, contributing to a spectrum of cardiovascular abnormalities and increasing the risk of adverse cardiovascular events. Untreated severe OSA roughly doubles the incidence of heart failure and increases the risk of stroke by about 60% [2]. Chronic intermittent hypoxia, heightened sympathetic tone, systemic inflammation and oxidative stress trigger a cascade of structural and electrical remodeling that manifests as hypertension [3, 4], arrhythmias [5, 6], heart failure with preserved or reduced ejection fraction [7, 8], pulmonary hypertension and right-sided heart disease [9, 10], coronary artery disease [11], even sudden cardiac death [12, 13]. Cardiac magnetic resonance imaging (CMR) is unrivaled in its ability to precisely quantify cardiac dimensions, global and regional function, myocardial mechanics, and tissue characterization and myocardium fibrosis. Several CMR studies have examined cardiac alterations in OSA. Curta et al. [14].analyzed 4978 participants from the UK Biobank and reported that OSA exerts sex‑specific remodeling: in women, both left and right ventricular end‑diastolic volume indices (LVEDVI and RVEDVI) were significantly lower, whereas in men the dominant finding was a higher right ventricular ejection fraction (RVEF). Li et al. [15].compared 60 OSA patients with 20 healthy controls using contrast‑enhanced CMR and found that absolute LV mass was elevated in severe OSA, and global circumferential strain (GCS) was decreased indicating early subclinical LV impairment while LVEF remained preserved. OSA also induces RV remodeling, including RV hypertrophy and impaired longitudinal strain, which may progress to RV dysfunction in severe cases [16]. Additionally, atrial fibrillation in OSA patients has been linked to atrial fibrosis detected by late gadolinium enhancement on CMR, potentially serving as a substrate for arrhythmogenesis [17]. These findings underscore the importance of early detection and intervention in OSA to mitigate adverse cardiac outcomes.

Notably, these studies primarily focused on conventional volumetric and functional parameters such as LVEF, LVEDV, neither included advanced parametric mapping (T1 or T2) for tissue characterization, nor did they assess right ventricular strain comprehensively or evaluate remodeling indices such as right ventricular end‑systolic remodeling index (RVESRI) [18]. Thus, the aim of this study is to analyze biventricular function, strain, right ventricular remodel indices, as well as ventricular blood pool T1 and T2 value in patients with severe versus non‑severe OSA, and to explore their correlations with Apnea- Hypopnea Index (AHI) and Oxygen Desaturation Index (ODI).

Methods

Study design and population

This study was conducted in accordance with the Declaration of Helsinki (revised in 2013) and approved by the Institutional Ethics Committee of China-Japan Friendship Hospital (2022-KY-226-1). Written informed consent was obtained from each patient prior to CMR imaging.

This cross-sectional observational study prospectively enrolled patients with OSA who underwent full-night polysomnography (PSG) at the Sleep Center of our hospital between March 2024 and March 2025. The inclusion criteria were as follows: (I) patients aged between 20 and 65 years; (II) participants completed CMR imaging within 24 h following PSG. (III) patients with normal findings on chest low-dose computed tomography. The exclusion criteria included: (I) Patients who received treatment before CMR; (II) patients who refused or were unable to tolerate CMR; (III) patients with poor CMR image quality that precluded the evaluation of cardiac function and strain; (IV) patients with arrhythmias; (V) patients with diabetes, coronary heart disease, congenital heart disease, cardiomyopathy, or pulmonary embolism. (VI) patients who had undergone coronary artery bypass grafting, stent implantation, left atrial appendage occlusion or inferior vena cava filter implantation; Fig. 1 illustrates the study flowchart.

Fig. 1.

Flow chart of this study

Polysomnography examination and assessment

All patients underwent full-night PSG examination using the Nox T3 (Nox Medical, Iceland). Six-channel electroencephalography, electrooculography, submental electromyography, and electrocardiography were recorded using surface electrodes.

Chest and abdominal belts were used to record respiratory movements. Airflow was monitored with an oronasal transducer, and arterial oxygen saturation was measured using finger pulse oximetry. All PSG scoring was performed by experienced PSG technicians who each had at least 10 years of continuous experience in manual scoring according to the standard criteria of American Academy of Sleep Medicine for the Scoring of Sleep and Associated Events [19]. The senior reference scorer was a board-certified sleep medicine physician with over 10 years of experience. Apnea- Hypopnea Index (AHI) was treated as the primary severity measure. AHI is calculated as the average number of apneas (complete cessation of airflow for ≥ 10 s) plus hypopneas (≥ 30% reduction in airflow lasting ≥ 10 s accompanied by a ≥ 3% oxygen desaturation or an arousal) per hour of sleep. The Oxygen Desaturation Index (ODI) was defined as the average number of times per hour of sleep that arterial oxygen saturation dropped by ≥ 3% from baseline. OSA was defined as an AHI ≥ 5 events/hour; Further, the severity of OSA was defined as follows: mild OSA: 5 ≤ AHI < 15 events/hour; moderate OSA: 15 ≤ AHI < 30 events/hour; severe OSA: AHI ≥ 30 events/hour [20]. Oxygen Desaturation Index (ODI) is calculated as the average number of times per hour of sleep that arterial oxygen saturation drops by ≥ 3% from baseline. ODI specifically captures the hypoxemic burden of respiratory events, independent of whether the event is classified as apnea or hypopnea. Apnea Index (AI) is the average number of apneic events (complete airflow cessation) per hour of sleep. AI reflects the component of sleep-disordered breathing characterized by complete rather than partial airway obstruction. Hypopnea Index (HI) is the average number of hypopneic events (partial airflow reduction) per hour of sleep. HI captures the burden of events with persistent but reduced airflow, which may be associated with less severe oxygen desaturation compared to apneas. AHI, ODI, AI, and HI each capture different physiological dimensions of sleep-disordered breathing.

Cardiac magnetic imaging protocol

All patients underwent non-contrast CMR on a 1.5 Tesla MR scanner (MAGNETOM Aera, Siemens Healthcare, Germany) in a supine position with an 18-channel body coil within 24 h after PSG examination and images were acquired during end-expiratory breath holds with retrospective electrocardiographic(ECG) gating without contrast. Retrospective ECG gating records the patient’s ECG signal during the entire cardiac cycle and sorts the acquired k‑space data into user‑defined cardiac phases (e.g., 25 phases per R‑R interval). This approach allows reconstruction of cine images across the full cardiac cycle with high temporal resolution and is particularly useful for assessing both systolic and diastolic function without missing end‑systolic or end‑diastolic frames. Initially, scout images were acquired to adjust the long and short axis of the heart. Then, three cine long-axis images in two-, three- and four-chamber view and 5–7 cine Short-axis images from apex to basal segment were acquired with a retrospectively ECG-gated segmented k-space bSSFP pulse sequence (TrueFISP) with the following parameters: repetition time(TR) = 34.5ms, echo time(TE) = 1.1 ms, Flip angle = 50–60°, slice thickness = 6 mm, in-plane spatial resolution = 1.8 × 1.8 mm², temporal resolution = 40 ms, and 25 reconstructed cardiac phases.

T1-mapping was obtained with a modified Look-Locker inversion-recovery (MOLLI) sequence with a single-shot balanced steady-state free precession (bSSFP) readout including the short-axis slices and long-axis four-chamber slice with the following parameters: TR = 1.14ms, TE = 280.56 ms, Flip angle = 35°, Voxel size = 1.4 × 1.4 × 8 mm³, Field of view = 360 × 256 mm, GRAPPA = 2, 24 reference lines, cardiac delay time TD = 435 ms, and phase partial Fourier 7/8. Each slice was 8-mm thick with a 2-mm gap. 2 inversions pulses with 5 and 3 images were acquired after each inversion pulse respectively and 3 heart beats to recover before the 2nd inversion If necessary, shimming and center frequency adjustments were performed to generate images free from off-resonance artifacts.

T2-mapping was performed using a true steady-state precession fast imaging sequence (TrueFISP T2 Map) as T2 Map using the following parameters: TR = 1.06ms, TE = 193.27 ms, Flip angle = 70°, Slice thickness = 8.0 mm, voxel = 1.9 × 1.9 × 8.0 mm³. Field of view = 360 × 256 mm, GRAPPA = 2, T2 preparation: 0 ms, 25 ms, 55 ms, cardiac delay time TD = 517 ms, phase partial Fourier 7/8. Each slice was 8-mm thick with a 2-mm gap.

CMR analysis

CMR analysis was performed by two cardiovascular radiologists with 10 years of experience in cardiovascular imaging, both of whom were blinded to all clinical data including PSG results and OSA severity classification. Left and right ventricular function were measured on the short-axis and long-axial 4-chamber cine images with the cvi 42 software (Circle Cardiovascular Imaging Inc., Calgary, AB, Canada). Right ventricular remodeling index including right ventricular end-systolic remodeling index (RVESRI) [18] and right ventricular end-diastolic remodeling index (RVEDRI) was respectively measured on the long-axis 4-chamber cine image of CMR at end-diastole and end-systole(Fig. 2) on SyngoVia workstation (Siemens Healthcare, Erlangen, Germany). They are defined as the ratio of the RV lateral free wall length to the interventricular septal length at end-systole and end-diastole, reflects the adequacy of RV adaptation to afterload [18].

Fig. 2.

Measurement of right ventricular end-diastolic remodeling index (RVEDRI) and right ventricular end-systolic remodeling index (RVESRI) on four-chamber cine images. (A) End-diastole: RVEDRI is calculated as the ratio of the RV lateral free wall length (green line) to the interventricular septal length (yellow line) at end-diastole. (B) End-systole: RVESRI is calculated as the ratio of the RV lateral free wall length (green line) to the interventricular septal length (yellow line) at end-systole. These indices reflect the geometric adaptation of the right ventricle to loading conditions: RVEDRI represents diastolic configuration, while RVESRI integrates systolic shortening and septal interaction, serving as a marker of RV systolic adaptation to afterload

Biventricular mechanics analysis were performed using the Cvi 42 software suite (Circle Cardiovascular Imaging Inc., Calgary, AB, Canada) including the left and right ventricular global longitudinal strain (LVGLS, RVGLS) and GLS rate (LVGLSR, RVGLSR) were obtained from the two-, three-, and four-chamber views. The left and right global circumferential strain (LVGCS, RVGCS), GCS rate (LVGCSR, RVGCSR), and global radial strain (LVGRS, RVGCS) and GRS rate (LVGRSR, RVGRSR) analyses were derived by sketching endocardial and epicardial borders on the basal, middle, and apical planes of the short-axis view.

According to Deng et al. [21], right and left ventricular blood pool T1 (RVT1, LVT1) and T2 values (RVT2, LVT2) were respectively measured on the four-chamber slice of T1 map and T2 map on SyngoVia workstation(Siemens Healthcare Sector, Forchheim, Germany). Region of interest (ROI) of the left and right ventricular blood pool were delineated along the endomyocardium on the four-chamber T1 Map and T2 Map (Fig. 3).

Fig. 3.

Measurement of right and left ventricular blood pool T1 (A) and T2 (B) on a four-chamber T1 map and T2 map. ROIs were traced along the endocardial border, with careful manual adjustment to exclude myocardium, trabeculations, and papillary muscles. For the right ventricle, the ROI was slightly indented from the endocardial border in areas where the myocardium was thin or trabeculations were present

Statistical analysis

Statistical analysis was performed using SPSS v26.0 (SPSSInc, Chicago, IL, USA) and MedCalc ^®Statistical Software version 20.211(MedCalc Software Ltd, Ostend,

Belgium; https://www.medcalc.org; 2023). Normally distributed measurement data were described as mean ± standard deviation (SD). Non-normally distributed data were described by the median and interquartile range (IQR). Given the limited sample size of mild OSA cases (n = 4), separate analysis for this subgroup was not feasible. Therefore, mild and moderate cases were combined into a single ‘non-severe’ group, which for interpretive purposes should be considered a predominantly moderate-OSA reference cohort. CMR functional and tissue characters parameters between non-severe and severe OSA groups were compared by t-tests or Mann-Whitney U tests as appropriate. For all primary group comparisons, effect sizes were calculated using Cohen’s d (mean difference divided by pooled standard deviation), with values of 0.2, 0.5, and 0.8 interpreted as small, medium, and large effects, respectively. To evaluate whether the observed differences in CMR parameters between severe and non‑severe OSA groups were independent of obesity, we performed multivariable linear regression for each CMR parameter that showed significant between‑group differences. Models included AHI and ODI in sensitivity analyses and BMI as independent variables. Furtherly, we have performed ROC analysis to compare the discriminatory performance of RVT2 alone versus RVT2/LVT2 ratio for distinguishing severe from non-severe OSA. Spearman’s rank correlation analysis was performed and demonstrated with Correlation heatmap between CMR parameters and AHI/ODI in OSA patients. Differences were considered statistically significant at two-sided P values < 0.05.

Results

Demographic characteristics

A total of 75 patients (62 males, mean age = 43.7 ± 9.4 years) were included in this study. There were 4 mild, 17 moderate cases (12 males, mean age 45.3 ± 10.7 years) in non-severe group, and 54 cases (47 males, mean age 43.5 ± 8.9 years) in the severe group. Table 1 shows the baseline clinical data of all patients. There were no statistically significant differences in age, gender, height, heart rate, systolic blood pressure, diastolic blood pressure, and smoker between the non-severe group and the severe group (P > 0.05). There were differences in body weight, body mass index (BMI), and body surface area (BSA) (P < 0.05). Compared with the non-severe group, the severe group had higher AHI, Oxygen Desaturation Index(ODI), Apnea Index(AI), Hypopnea Index (HI) and Time spent with SpO2 < 90% (T90), and lower baseline oxygen saturation (SPO_{2 baseline}), mean oxygen saturation (SPO_{2 mean}) as well as nadir oxygen saturation (SPO_{2 nadir}).

Table 1.

Demographic characteristics of patients with OSA

characteristics	OSA
	Total	Non-severe		Severe		t/χ²/U	p-value
Case number	75	21		54
Age(years)	43.7 ± 9.3	44.1 ± 10.7		43.5 ± 8.9		0.272	0.787
Sex(Male/Female)	62/13	15(71.4%)/6(28.6%)		47(87.0%)/7(13.0%)		2.571	0.109
Height (m)	1.7 ± 0.1	1.7 ± 0.1		1.7 ± 0.0		-0.497	0.62
Body Weight(kg)	86.4 ± 14.1	80.1 ± 11.8		88.9 ± 14.1		-2.525	0.014*
BMI(kg/m2)	29.0 ± 3.4	27.1 ± 2.8		29.7 ± 3.3		-3.174	0.002*
BSA(m2)	2.0 ± 0.1	1.9 ± 0.1		2.0 ± 0.2		-2.065	0.042*
Heart Rate(bpm)	73.4 ± 9.7	70.8 ± 10.5		74.4 ± 9.7		-1.316	0.193
SBP(mmHg)	127.4 ± 12.5	126.3 ± 12.2		128.2 ± 12.8		-0.544	0.588
DBP(mmHg)	87.3 ± 12.4	84.7 ± 12.7		88.6 ± 12.3		-1.127	0.263
Smokers(n,%)	36(48%)	8(38.1%)		28(51.9%)		1.146	0.284
Polysomnography parameters
AHI (events/hr)	47.4(23.8,69.0)	12.7(10.3,19.9)		58.2(43.3,77.1)		1134	< 0.001*
ODI (events/hr)	55.7(16.1,77.1)	7.9(5.9,15.4)		66.6(36.0,80.3)		1127	< 0.001*
AI (events/hr)	35.4(6.8,54.8)	1.8(0.5,5.5)		54.5(29.8,71.9)		1110	< 0.001*
HI (events/hr)	9.8(4.6,15.3)	10.7(9.1,13.2)		9.3(3.5,17.8)		473.5	0.27
SpO2baseline(%)	96.0(95.0,97.0)	96.0(96.0,97.0)		95.0(94.0,97.0)		377.5	0.022*
SpO2mean(%)	94.0(90.0,96.0)	96.0(96.0,96.5)		91.0(88.0,94.3)		117	< 0.001*
SpO2nadir(%)	72.0(57.0,84.0)	87.0(84.0,90.0)		65.0(53.0,76.3)		52.5	< 0.001*
T90(%)	13.7(0.7,42.5)	0.2(0.2,0.8)		29.0(7.2,50.5)		856.5	< 0.001*

Data are presented as mean ± SD for normally distributed continuous variables, median (IQR) for non-normally distributed continuous variables, and n (%) for categorical variables; BMI: body mass index; BSA: Body Surface Area; SBP: systolic blood pressure; DBP: diastolic blood pressure; AHI: apnea-hypopnea index; ODI: oxygen desaturation index; AI: Apnea Index; HI: Hypopnea Index; SPO2 baseline: baseline oxygen saturation, SPO2 mean: mean oxygen saturation; SPO2 nadir: nadir oxygen saturation. T90%:Time spent with SpO2 < 90%.*p < 0.05

Comparison of non-contrast CMR parameters between non-severe and severe group

Left and right ventricular functional parameters are presented in Table 2. Except that the left ventricular myocardial mass (LVM) in the severe group was higher than that in the non-severe group (131.12 ± 25.42 vs. 115.57 ± 29.59; p = 0.038, Cohen’s d = 0.57), there were no statistically significant differences in the other conventional functional parameters between two groups. Notably, RVESRI was significantly higher in the severe group compared to the non-severe group (1.37 ± 0.11 vs. 1.28 ± 0.08; p = 0.002, Cohen’s d = 0.96), whereas RVEDRI was similar between two groups (1.52 ± 0.11 vs. 1.53 ± 0.10; p = 0.817, Cohen’s d = 0.10).

Table 2.

Comparison of biventricular function between severe and non-severe OSA groups

biventricular function	OSA group
	Non-severe (n = 21)	Severe (n = 54)	t	p-value	Cohen’s d
Left Ventricle
LVEF(%)	67.88 ± 4.62	65.74 ± 6.17	1.316	0.193	0.39
LVEDV(ml)	112 ± 38.00	117 ± 34.50	-0.865	0.332	0.14
LVESV(ml)	37.64 ± 11.75	41.90 ± 12.04	-1.279	0.206	0.36
LVSV(ml)	76.00 ± 19.50	77.00 ± 17.75	0.809	0.418	0.05
LVCO/(L/min)	5.38 ± 0.81	5.91 ± 1.34	-1.528	0.131	0.47
LVM(g)	115.57 ± 29.59	131.12 ± 25.42	-2.113	0.038*	0.57
LVEDVI(ml/m2)	61.00 ± 9.50	57.50 ± 12.00	-0.438	0.661	0.32
LVESVI (ml/m2)	19.54 ± 4.65	20.50 ± 5.44	-0.656	0.514	0.19
LVSVI (ml/m2)	40.85 ± 6.12	38.91 ± 4.79	1.356	0.179	0.35
LVCI (L/min/ m2)	2.83 ± 0.25	2.87 ± 0.50	-0.443	0.749	0.10
LVMI (g/ m2)	59.58 ± 11.05	64.56 ± 10.11	-1.733	0.088	0.47
Right Ventricle
RVEF(%)	55.80 ± 4.89	57.23 ± 5.74	-0.924	0.359	0.27
RVEDV(ml)	91.20 ± 37.29	91.47 ± 50.96	-0.135	0.893	0.01
RVESV(ml)	43.48 ± 22.79	39.69 ± 23.36	-0.330	0.741	0.16
RVSV(ml)	54.17 ± 19.86	54.06 ± 27.57	0.148	0.882	0.00
RVCO(L/min)	3.74 ± 2.05	3.88 ± 2.36	0.593	0.553	0.06
RVEDVI/(ml/ m2)	49.30 ± 21.33	49.66 ± 18.64	-0.633	0.527	0.02
RVESVI/(ml/ m2)	22.85 ± 9.68	20.17 ± 10.21	-0.835	0.403	0.27
RVSVI/(ml/ m2)	28.31 ± 13.88	26.41 ± 11.52	-0.472	0.637	0.15
RVCI/(L/min/ m2)	2.07 ± 1.15	1.92 ± 1.10	0.263	0.793	0.13
RVEDRI	1.53 ± 0.10	1.52 ± 0.11	0.232	0.817	0.10
RVESRI	1.28 ± 0.08	1.37 ± 0.11	3.241	0.002*	0.96

*p < 0.05; Cohen’s d effect size interpretation: 0.2 = small, 0.5 = medium, 0.8 = large

LVEF: left ventricular eject fraction; LVEDV: left ventricular end-diastolic volume; LVESV: left ventricular end-systolic volume; LVSV: left ventricular stroke volume; LVCO: left ventricular cardiac output; LVM: left ventricular mass; LVEDVI: left ventricular end-diastolic volume index; LVEDVI: left ventricular end-systolic volume index; LVSVI: left ventricular stroke volume index༛LVCI: left ventricular cardiac index༛LVMI: left ventricular mass index; RVEF: right ventricular eject fraction; RVEDV: right ventricular end-diastolic volume; RVESV: right ventricular end-systolic volume; RVSV: right ventricular stroke volume༛ RVCO: right ventricular cardiac output; RVEDVI: right ventricular end-diastolic volume index; RVEDVI: right ventricular end-systolic volume index; RVSVI༚right ventricular stroke volume index༛RVCI: right ventricular cardiac index༛

Table 3 demonstrates the left and right ventricular strain and strain rate. Compared with the non-severe group, the absolute value of left ventricular global radial strain (LVGRS) was significantly lower in the severe group (28.27 ± 5.06% vs. 33.85 ± 5.65%; p < 0.001, Cohen’s d = 1.04), as was left ventricular global circumferential strain (LVGCS) (-17.68 ± 2.18% vs. -19.69 ± 2.02%; p < 0.001, Cohen’s d = 0.96) and circumferential strain rate (LVGCSR) (-0.99 ± 0.15 vs. -1.06 ± 0.14; p = 0.029, Cohen’s d = 0.49). However, left ventricular longitudinal strain (LVGLS), left longitudinal strain rate (LVGLSR), and left ventricular global radial strain rate (LVGRSR) were comparable between two groups (all p > 0.05). Notably, all right ventricular strain parameters were significant lower in the severe group with large effect sizes. The absolute values of right ventricular global longitudinal strain (RVGLS) were lower in severe OSA (-22.84 ± 3.19% vs. -25.57 ± 2.89%; p = 0.007, Cohen’s d = 0.89), as were RV global circumferential strain (RVGCS) (-12.13 ± 2.83% vs. -15.09 ± 1.72%; p < 0.001, Cohen’s d = 1.25) and RV global radial strain (RVGRS) (20.10 ± 4.96% vs. 23.30 ± 4.50%; p = 0.001, Cohen’s d = 0.67). All right ventricular strain rates also were significant lower with effect sizes ranging from moderate to large (Cohen’s d range: 0.65–0.94).

Table 3.

Comparison of biventricular Strain between severe and non-severe OSA groups

Strain analysis	OSA group
	Non-severe (n = 21)	Severe (n = 54)	t	p-value	Cohen’s d
Left Ventricle
LVGLS(%)	-18.77 ± 2.13	-17.93 ± 2.02	-1.591	0.116	0.40
LVGRS(%)	33.85 ± 5.65	28.27 ± 5.06	4.146	< 0.001*	1.04
LVGCS(%)	-19.69 ± 2.02	-17.68 ± 2.18	-3.674	< 0.001*	0.96
LVGLSR(1/s)	-0.93 ± 0.09	-0.96 ± 0.17	0.740	0.462	0.22
LVGRSR(1/s)	1.78 ± 0.26	1.62 ± 0.39	1.682	0.097	0.48
LVGCSR(1/s)	-1.06 ± 0.14	-0.99 ± 0.15	-2.222	0.029*	0.49
Right Ventricle
RVGLS(%)	-25.57 ± 2.89	-22.84 ± 3.19	-3.408	0.007*	0.89
RVGRS(%)	23.30 ± 4.50	20.10 ± 4.96	3.642	0.001*	0.67
RVGCS(%)	-15.09 ± 1.72	-12.1261 ± 2.83	-3.756	< 0.001*	1.25
RVGLSR(1/s)	-1.59 ± 0.85	-1.07 ± 0.80	-2.451	0.017*	0.63
RVGRSR(1/s)	1.50 ± 0.33	1.18 ± 0.36	3.493	0.001*	0.94
RVGCSR(1/s)	-0.99 ± 0.26	-0.80 ± 0.21	-3.379	0.001*	0.80

Data are presented as mean ± standard deviation; *p < 0.05; Cohen’s d effect size interpretation: 0.2 = small, 0.5 = medium, 0.8 = large; LVGLS: left ventricular global longitudinal strain; LVGRS: left ventricular global radial strain; LVGCS: left ventricular global circumferential stain; LVGLSR: left ventricular global longitudinal strain rate; LVGRSR: left ventricular global radial strain rate; LVGCSR: left ventricular global circumferential stain rate; RVGLS: right ventricular global longitudinal strain; RVGRS: right ventricular global radial strain; RVGCS: right ventricular global circumferential stain; RVGLSR: right ventricular global longitudinal strain rate; RVGRSR: right ventricular global radial strain rate; RVGCSR: right ventricular global circumferential stain rate

Left and right ventricular blood pool T1 and T2 value are summarized in Table 4. No significant differences were observed in RVT1, LVT1 or the RVT1/LVT1 ratio between the non-severe and severe groups. In contrast, the severe group exhibited significantly lower RVT2 and RVT2/LVT2 ratio compared to the non-severe group (p < 0.05), while LVT2 remained unchanged. A sensitivity analysis (Supplementary Table S1) comparing our original tracing method to a small central ROI in 30 randomly selected patients demonstrate the absolute values were slightly higher with the conservative method (by ~ 1‑2 ms, consistent with reduced partial volume averaging with low‑T2 myocardium), but the differences between severe and non‑severe groups and the correlation coefficients with AHI were nearly identical (original ROI: ρ = − 0.431; conservative ROI: ρ= − 0.428). This indicates that while partial volume effects may introduce a small systematic offset, they do not materially affect the study’s main conclusions regarding group differences or associations with OSA severity. Furthermore, RVT2 weakly-to-moderately correlated with SPO_{2 baseline} (ρ = 0.293), SPO_{2 mean} (ρ = 0.418), SPO_{2 nadir} (ρ = 0.294), AHI(ρ=-0.435) and ODI(ρ=-0.425)(Supplementary Figure S1).

Table 4.

The Right and left ventricular blood pool T1 and T2 between severe and non-severe OSA groups

T1 and T2 value	OSA group
	Non-severe (n = 21)	Severe (n-54)	t	P-value	Cohen’s d
RVT2(ms)	141.41 ± 12.04	126.95 ± 11.40	4.490	<0.001*	1.23
LVT2(ms)	159.88 ± 15.50	153.59 ± 17.12	1.349	0.182	0.38
RVT2/LVT2 ratio	0.88 ± 0.03	0.83 ± 0.07	3.873	<0.001*	0.93
RVT1(ms)	1487.64 ± 75.71	1449.66 ± 85.03	-1.646	0.104	0.47
LVT1(ms)	1522.00 ± 78.00	1552.70 ± 85.45	-0.849	0.396	0.37
RVT1/LVT1 ratio	0.95 ± 0.02	0.95 ± 0.04	-1.024	0.306	0.00

Data are presented as mean ± standard deviation. *p < 0.05; Cohen’s d effect size interpretation: 0.2 = small, 0.5 = medium, 0.8 = large

RVT2: right ventricular blood pool T2 value; LVT2: ventricular blood pool T2 value; RVT1: right ventricular blood pool T1 value; LVT1: left ventricular blood pool T1 value

After adjusting for BMI in multivariable linear regression, AHI remained associated with LVM (β = 0.21, p = 0.024), LVGCS (β = − 0.056, p = 0.004), LVGRS (β = − 0.19, p = 0.008), RVGLS (β = − 0.11, p = 0.009), RVGCS (β = − 0.17, p < 0.001), RVGRS (β = − 0.14, p = 0.021), RVESRI (β = 0.003, p = 0.015), and RVT2 (β = − 0.24, p = 0.003). BMI was independently associated only with LVM (β = 2.15, p < 0.001), but not with any strain parameter, RVESRI, or RVT2 (all p > 0.05, Supplementary Table S2).

Similar, After adjusting for BMI in multivariable linear regression, ODI remained associated with LVM (β = 0.19, p = 0.018), LVGCS (β = − 0.052, p = 0.005), LVGRS (β = − 0.17, p = 0.007), RVGLS (β = − 0.10, p = 0.01), RVGCS (β = − 0.16, p < 0.001), RVGRS (β = − 0.13, p = 0.022), RVESRI (β = 0.003, p = 0.017), and RVT2 (β = − 0.22, p = 0.002). BMI was independently associated only with LVM (β = 2.18, p < 0.001), but not with any strain parameter, RVESRI, or RVT2 (all p > 0.05, Supplementary Table S3).

ROC analysis for discriminating severe from non-severe OSA showed an AUC of 0.758 (95% CI: 0.646–0.850) for RVT2 alone and an AUC of 0.695 (95% CI: 0.578–0.796) for the RVT2/LVT2 ratio. The difference was not significant (DeLong test, Z = 0.779,P = 0.4357) (Fig. 4).

Fig. 4.

ROC analysis for discriminating severe from non‑severe OSA. ROCs showed an AUC of 0.758 (95% CI: 0.646–0.850) for RVT2 alone and an AUC of 0.695 (95% CI: 0.578–0.796) for the RVT2/LVT2 ratio. The difference was not significant (DeLong test, Z = 0.779, p = 0.4357), indicating that the ratio does not provide incremental diagnostic value over RVT2 alone in this setting

Correlation of CMR parameters and severity of OSA

The Correlation Heatmap (Fig. 5) shows LVM(ρ = 0.363, P = 0.0014), LVGLS(ρ = 0.313, P = 0.006), and LVGCS(ρ = 0.448, P = 0.0001) weakly to moderately correlated with AHI, whereas LVGRS showed an moderate inverse relationship (ρ = − 0.441, P = 0.0001). Comparable patterns also emerged for ODI: LVM (ρ = 0.355, P = 0.002), LVGLS (ρ = 0.322, P = 0.005), and LVGCS (ρ = 0.434, P = 0.0001) were positively correlated, whereas LVGRS (ρ=–0.426, P = 0.0001), LVGCSR(ρ=-0.325,P = 0.004), and LVGRSR(ρ=–0.325, P = 0.004) exhibited weak to moderate associations.

Fig. 5.

Correlation heatmap of left ventricular strain, blood pool T1, T2 values and AHI, ODI. Color intensity represents the strength of Spearman’s correlation coefficient (ρ), with red indicating positive correlations and blue indicating negative correlations. AHI=Apnea-hypopnea index; ODI=oxygen desaturation index

The Correlation Heatmap (Fig. 6) demonstrates RVESRI (ρ = 0.307, P = 0.007), RVGLS (ρ = 0.335, P = 0.003),RVGCS (ρ = 0.476, P < 0.001) increased with AHI, while RVGRS (ρ =-0.397, P = 0.0004), RVGRSR (ρ =-0.301, P = 0.0087), RT2 (ρ =-0.435, P = 0.0001) showed an inverse relationship. Comparable correlations also were found with ODI: RVGLS (ρ = 0.325, P = 0.004), RVGCS (ρ = 0.459, P < 0.0001) were positively correlated, whereas RVGRS (ρ = − 0.395, P = 0.0004) and RVT2(ρ=–0.425, P = 0.0001) exhibited weak to moderate associations.

Fig. 6.

Correlation heatmap of right ventricular strain, blood pool T1, T2 values and AHI, ODI. Color intensity represents the strength of Spearman’s correlation coefficient (ρ), with red indicating positive correlations and blue indicating negative correlations. AHI=Apnea-hypopnea index; ODI=oxygen desaturation index

Exploratory sex-subgroup analyses

Given the predominance of male participants in this cohort (62/75, 82.7%), we performed exploratory subgroup analyses stratified by sex for the key CMR parameters that differed significantly between severe and non‑severe OSA groups. Due to the small number of female patients (n = 13 overall; n = 6 non‑severe, n = 7 severe), formal statistical comparisons were not performed, and no definitive sex‑specific conclusions can be drawn(Supplementary Table S4).

Discussion

Our study was designed to compare CMR functional, remodeling parameters and ventricular blood pool T1 and T2 mapping between severe and non‑severe OSA patients, and to explore whether these parameters correlate with AHI and ODI. There are several important findings: (I) subclinical cardiac dysfunction detectable through strain, remodeling indices, and blood pool T2 mapping may precede abnormalities in conventional CMR parameters; (II) Compared with non-severe group, the severe OSA group exhibited lower strain and higher RVESRI; and (III) RV blood pool T2 (RVT2) may serve as a non-invasive biomarker strongly linked to nocturnal hypoxemia and cardiopulmonary stress.

Since four mild OSA patients entered in this cohort, we were unable to evaluate myocardial abnormalities in patients with mild OSA, therefore, we combined mild and moderate cases into a non-severe OSA group. To address the absence of a concurrent healthy control group, we compared our non-severe OSA group’s CMR parameters with published normative reference values derived from healthy cohorts [21–24]. The non-severe group showed LVGCS (–19.7 ± 2.0% vs. normative − 19.9 ± 2.5%) [22], RVGLS (–25.6 ± 2.9% vs. − 25.1 ± 3.3%) [23], and LVMI (59.6 ± 11.1 g/m² vs. normal range 54–66 g/m²) [24]that were broadly within normal limits. Regarding RVT2, no large normative study exists, but our non-severe group’s RVT2 (141 ± 12 ms) is similar to values reported in healthy controls by Deng et al. [21]. These comparisons suggest that the non-severe group did not exhibit substantial subclinical dysfunction.

Curta et al. [14]. reported that OSA is associated with sex‑specific changes in conventional CMR parameters: lower LVEDVI and RVEDVI in females, and higher RVEF in males. They did not assess myocardial strain, ventricular blood pool parametric T1/T2 mapping, or RVESRI, RVEDRI remodeling indices. Our findings complement rather than contradict these observations. While we observed no significant difference in RVEF between severe and non‑severe OSA, we found widespread impairment of RV strain (RVGLS, RVGCS, RVGRS) that correlated with AHI and ODI. This suggests that strain abnormalities may precede overt ejection fraction decline in the pathophysiology of OSA‑related right heart disease. Similarly, the lower RVT2 in severe OSA and its correlation with hypoxemic burden represent novel observations not previously reported. Thus, the two studies address different layers of cardiac assessment: Curta et al. examined macroscopic changes using conventional cine‑based parameters in a large cohort [14], but we examine subclinical mechanical and tissue‑level changes using advanced parameters in a smaller, more severely affected cohort. Both are important for a complete understanding of OSA‑related cardiac injury.

Compared to non-severe OSA group, the severe group exhibited higher LVM and lower left ventricular global circumferential and radial strain. These findings align with previous reports linking OSA to LV hypertrophy and early systolic dysfunction [15, 25, 26], likely driven by chronic intermittent hypoxia, sympathetic overactivation, and increased afterload due to hypertension. The correlations between LV strain parameters (LVGCS, LVGRS), and both AHI and ODI further support a severity-dependent pattern of LV remodeling in OSA. Speckle tracking echocardiography showed RV dysfunction in OSA [27, 28]. Li et al. reported the conventional RV functional parameter like RVEF, RVEDV, RVESV was similar between non-severe and severe groups [15]. Our study reveals widespread impairment in RV longitudinal, circumferential, and radial strain in severe OSA, more pronounced than in the LV, highlighting the RV’s greater vulnerability to OSA-related stress.

Although obesity may be a potential confounder, multivariable regression adjusting for BMI demonstrated that AHI remained independently associated with all key CMR parameters except LVM where both AHI and BMI were independent predictors. To further address the concern regarding confounding by BMI, we furtherly performed sensitivity analyses using ODI as the primary exposure variable, the independent associations between ODI and all key CMR parameters remained significant after adjusting for BMI, with effect sizes and model R² values comparable to those obtained with AHI. These findings indicate that the observed cardiac alterations are linked to the hypoxemic burden of OSA rather than being an artifact of the specific severity index chosen.

RVESRI provided information on both size and systolic function in CTEPH patients [18], predicted outcomes in patients with pulmonary arterial hypertension [29]. To our knowledge, our study is the first to examine RV remodeling indices in OSA. RVESRI was significantly higher in severe OSA patients than in the non-severe OSA group, while RVEDRI was comparable between mild-to-moderate and severe groups. The possible mechanisms driving elevated RVESRI may be related to repetitive nocturnal hypoxemia triggers pulmonary vasoconstriction, increasing pulmonary vascular resistance and RV afterload [30, 31]. In contrast, the RV may maintain normal diastolic geometry (RVEDRI) through enhanced compliance or preload adaptation. OSA-induced intermittent hypoxia triggers neurohormonal activation, leading to fluid retention and increased preload. This could distend the RV in diastole, preserving its elliptical shape despite rising afterload. Moreover, despite nocturnal hypoxemia, daytime systemic oxygenation normalizes in many OSA patients. This may prevent chronic RV volume overload, maintaining baseline diastolic dimensions. Importantly, isolated elevation in RVESRI (without RVEDRI change) may signal early RV systolic maladaptation to OSA severity, preceding diastolic structural changes. Whether a preserved RVEDRI indicating normal diastolic geometry implies that systolic dysfunction is reversible with treatment before permanent structural remodeling occurs remains entirely speculative. No longitudinal interventional studies have yet examined whether changes in RVESRI or RVEDRI are reversible with OSA therapy. Therefore, this hypothesis requires dedicated prospective studies with serial CMR assessments before and after treatment, ideally with long‑term follow‑up to determine whether normalization of RVESRI and preservation of RVEDRI predict favorable clinical outcomes. In this study, we observed that AHI and ODI were both significantly elevated in severe OSA, but ODI demonstrated slightly stronger correlations with RV strain parameters and RVT2, consistent with a mechanistic link between hypoxemic burden and the cardiac alterations we identified.

T1 and T2 mapping offer non-invasive quantification of myocardial relaxation times, widely used for tissue characterization. While their application to myocardium is well established [32–34], use in blood pool analysis in OSA is novel. T2 relaxation time in blood is primarily determined by the magnetic susceptibility effect of deoxyhemoglobin, which creates microscopic field inhomogeneities that accelerate transverse relaxation. As oxyhemoglobin is diamagnetic and deoxyhemoglobin is paramagnetic, increased deoxyhemoglobin concentration shortens blood T2 in a concentration-dependent manner [35, 36] This principle underlies T2-based blood oximetry techniques validated in both animal models and human studies [37–43]. Deng et al. found it correlates with hemodynamics in pulmonary hypertension [21], while Emrich et al. demonstrated that RVT2/LVT2 ratio can identify left-to-right shunts [44].

In the context of OSA, repetitive nocturnal hypoxemia leads to chronically elevated tissue oxygen extraction and increased venous deoxyhemoglobin content, which is most directly reflected in the right ventricular blood pool, the recipient of systemic venous return before pulmonary reoxygenation. Therefore, the observed reduction in RVT2 in severe OSA likely represents cumulative venous hypoxemic burden rather than a direct measurement of arterial oxygenation. This interpretation is supported by the weak to moderate correlations we observed between RVT2 and nocturnal SpO₂ metrics, as well as with AHI and ODI.

In our cohort, RVT2 was significantly lower in severe OSA compared to non-severe OSA. This cross‑sectional association is consistent with the hypothesis that chronic intermittent nocturnal hypoxemia may lead to persistent alterations in venous blood composition, potentially reflected in awake T2 measurements. However, because CMR was performed during wakefulness when oxygen saturation is typically normalized, we cannot establish a direct temporal or causal link between acute desaturation events and the observed RVT2 values. Moreover, we acknowledge that blood pool T2 is influenced by multiple factors beyond oxygenation, including hematocrit, flow velocity, and partial volume effects from adjacent myocardium or trabeculations [42, 43]. Variations in hematocrit which affects blood water content and magnetic susceptibility can modulate T2 values independently of oxygen saturation. Additionally, turbulent or stagnant flow in the dilated or dysfunctional right ventricle may influence signal intensity and T2 measurements. Partial volume effects, particularly in the thin-walled RV, represent another potential source of variability. Unfortunately, hematocrit was not measured in this study, which limits our ability to disentangle the relative contributions of oxygenation and hematocrit to the observed RVT2 reduction. Future studies incorporating hematocrit measurements and advanced flow imaging are warranted to validate and refine the interpretation of RVT2 as a imaging marker in assessment of OSA.

Nevertheless, the inverse correlation between RVT2 and both AHI and ODI, along with its positive correlation with SpO₂ levels, suggest a weak to moderate association between RVT2 and hypoxemic burden in OSA. Moreover, its correlation with RV strain parameters suggests a mechanistic link between nocturnal desaturation and RV dysfunction. While these findings are promising, we caution against overinterpreting RVT2 as a direct measure of hypoxemia given the cross-sectional nature of our study and the multiple potential confounders discussed above. Rather, RVT2 may represent a composite marker of cardiopulmonary stress in OSA, integrating effects of hypoxemia, hemodynamic alterations, and potentially other unmeasured factors.

The RVT2/LVT2 ratio was also significantly lower in severe OSA, with an effect size slightly smaller than that of RVT2 alone. This raises the question of whether the ratio provides incremental value beyond RVT2 alone in OSA. Conceptually, the ratio may help normalize for individual variations in hematocrit, scanner calibration, and field inhomogeneities, as LVT2 (from the left ventricular pool) serves as an internal reference representing well-oxygenated blood. In conditions such as left-to-right shunts [43] or pulmonary hypertension [21], the ratio has demonstrated advantages. However, in our OSA cohort, where hypoxemia is predominantly systemic and venous, the right-sided blood pool is primarily affected while left-sided oxygenation remains relatively preserved during daytime scanning (as reflected by non-significant LVT2 changes). Thus, much of the signal resides in RVT2 itself. ROC analysis for discriminating severe from non-severe OSA gave an AUC of 0.758 for RVT2 alone versus 0.695 for RVT2/LVT2 ratio, suggesting no statistically significant incremental benefit. Therefore, while the ratio may be useful in certain clinical contexts, RVT2 alone appears sufficient for OSA severity stratification in this cohort.

Limitations

This study has several limitations that warrant consideration. First, a major limitation of this study is the striking under‑representation of women (13 of 75, 17.3%). While this proportion is comparable to the male predominance reported in many OSA cohorts, the accumulating evidence that OSA exerts sex‑specific cardiovascular effects [14, 45], seriously constrains the generalizability of our findings to the female population with OSA. The distinct cardiovascular risk profile in women with OSA characterized by greater susceptibility to endothelial dysfunction, HFpEF, and autonomic dysregulation suggests that women may experience a different pattern of cardiac remodeling than the predominantly RV‑strain‑driven changes observed in our male‑majority cohort. Importantly, beyond the small sample size, the female participants in our cohort were predominantly of child‑bearing age, but we did not collect data on reproductive status. This represents an additional limitation given that estrogen is known to exert protective cardiovascular effects against intermittent hypoxia‑induced injury. Our study therefore cannot address whether the observed RV strain impairment and RVESRI elevation in severe OSA are equally present in both pre‑ and post‑menopausal women, or whether any protective effects of endogenous estrogens attenuate these changes. Future studies must prioritize sex‑balanced enrollment to enable adequately powered sex‑stratified analyses. Only through such prospective, sex‑balanced, and hormonally informed designs can the generalizability of parameters such as RV strain and RVESRI which we identified as robust markers of OSA severity in males be properly validated for female populations. Second, while the exclusion of patients with significant comorbidities enhances internal validity by reducing confounding factors, it may reduce the applicability of our results to real-world OSA populations, who often present with multiple coexisting conditions. Third, we did not include a matched healthy control group which limited our ability to definitively establish baseline differences in CMR parameters between OSA and non-OSA individuals. No a priori statistical power analysis was performed to determine the required sample size for detecting the effects of interest. Consequently, while our overall sample size (n = 75) may be adequate for the primary two‑group comparisons (severe vs. non‑severe), the limited number of mild (n = 4) and moderate (n = 17) patients compromises our ability to detect early‑stage changes or to perform robust severity‑graded analyses. Future studies with balanced samples across mild, moderate, and severe strata are needed to define the trajectory of cardiac involvement from the earliest stages of OSA. Forth, we did not perform a formal statistical inter-rater reliability analysis for the PSG scoring in this specific study cohort. We acknowledge that formal reporting of inter-rater reliability metrics would strengthen methodological transparency. Fifth, both T2 and T2* are sensitive to deoxyhemoglobin concentration, future work could explore whether T2* mapping provides incremental sensitivity to hypoxemic burden in OSA patients. Finally, due to the cross-sectional design, causality cannot be established. Longitudinal studies are essential to validate these imaging biomarkers, and determine their prognostic value.

Conclusion

Compared with non-severe OSA group, patients with severe OSA exhibit significant biventricular alterations, including lower strain (particularly in the RV), adverse RV systolic remodeling, and reduced RV blood pool T2. Subclinical cardiac dysfunction detectable through strain, remodeling indices, and blood pool T2 mapping indicated that these metrics may serve as earlier markers of OSA-related cardiac involvement. Non-contrast CMR provides a robust and clinically feasible approach for risk stratification and early intervention in OSA.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

OSA

Obstructive sleep apnea

CMR

Cardiac Magnetic Resonance Imaging

AHI

Apnea-hypopnea index

ODI

Oxygen desaturation index

LV

Left ventricle

RV

Right ventricle

LVGRS

Left ventricular global radial strain

LVGCS

Left ventricular global circumferential strain

LVGLS

Left ventricular global longitudinal strain

RVGRS

Right ventricular global radial strain

RVGCS

Right ventricular global circumferential strain

RVGLS

Right ventricular global longitudinal strain

LVEDVI

Left ventricular end-diastolic volume index

RVEDVI

Right ventricular end-diastolic volume index

PSG

Polysomnography

ECG

Electrocardiogram

bSSFP

Balanced steady-state free precession

TR

Repetition time

TE

Echo time

MOLLI

Modified look locker inversion-recovery

GRAPAA

GeneRalized Autocalibrating Partial Parallel Acquisition

TD

Delay time

TrueFISP

True steady-state precession fast imaging sequence

RVESRI

Right ventricular end-systolic remodeling index

RVEDRI

Right ventricular end-diastolic remodeling index

ROI

Region of interest

SD

Standard deviation

IQR

Interquartile range

HI

Hypopnea Index

T90%

Time spent with SpO2 < 90%

SPO_{2 baseline}

Baseline oxygen saturation

SPO_{2 mean}

Mean oxygen saturation

SPO_{2 nadir}

Nadir oxygen saturation

LVM

Left ventricular myocardial mass

LVT1

Left ventricular blood pool T1 value

RVT1

Right ventricular blood pool T1 value

LVT2

Left ventricular blood pool T2 value

RVT2

Right ventricular blood pool T2 value

Author contributions

ML and XZ conceived and designed the study, supervised the project, and critically revised the manuscript. JW and ZC were responsible for data integrity, management, and the accuracy of the data analysis. JW, ZC, JD, AL, YN, HY, YG, and EM acquired the data. JW, ZC, JD, and EM analyzed and interpreted the data. JW, ZC, and JD drafted the manuscript. All authors read and approved the final manuscript.

Funding

This study was supported by National High Level Hospital Clinical Research Funding(2022-NHLHCRF-LX-01-0302) which support the design of the study.

Data availability

Processed data will be available upon request to the corresponding author of this work.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki (revised in 2013) and approved by the Institutional Ethics Committee of China-Japan Friendship Hospital (2022-KY-226-1). Written informed consent was obtained from each patient prior to CMR imaging.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.