- Academic Editor
†These authors contributed equally.
Background: Dilated cardiomyopathy (DCM) has a poor prognosis and high
mortality. The relationship between the deformation capacity of
the biatrial and biventricular regions in patients with DCM remains unclear.
Methods: This retrospective study used cardiovascular magnetic resonance
(CMR) to assess patient enrollment between September 2020 to May 2022. Feature
tracking (FT) was used to evaluate biventricular global radial strain (GRS),
global circumferential strain (GCS) and global longitudinal strain (GLS). Fast
long-axis method was used to evaluate biatrial GLS by analyzing balanced
steady-state free precession cine images. The median follow-up period was 362
days (interquartile range: 234 to 500 days). DCM patients were divided into two
groups based on the occurrence or non-occurrence of major adverse cardiac event
(MACE). The primary endpoint was defined as all-cause death, heart
transplantation, and adverse ventricular arrhythmia. The secondary end point
included hospitalizations due to heart failure. Cox regression analysis was
utilized for variables and Kaplan-Meier survival was utilized for clinical
outcomes. Results: There were 124 DCM patients (52.82
Dilated cardiomyopathy (DCM) remains a serious medical condition, and the challenges associated with risk stratification in DCM continue to pose ongoing challenges in clinical practice [1, 2]. Currently, cardiovascular magnetic resonance (CMR) is considered the gold standard for evaluating cardiac morphology, function, and tissue characterization [3]. Left ventricular (LV) ejection fraction (LVEF) has been the main criteria for evaluation of therapy in DCM [4]. However, studies have shown that improvement in LVEF following systemic therapy does not necessarily indicate recovery of systolic function [5, 6]. Furthermore, it has been established that left atrial (LA) volume and booster function are independent predictors of outcomes in DCM [7, 8], as atrial function is tightly coupled to ventricular relaxation and diastolic properties [9]. Studies have demonstrated that LA strain is a more sensitive measure than LA morphologic and functional alterations when reflecting LV diastolic dysfunction [10, 11].
Recent studies have highlighted the increasing potential of CMR in assessing myocardial deformation and its ability to predict clinical outcomes, surpassing traditional parameters [12, 13, 14]. Feature tracking (FT) has emerged as a useful technique for identifying subtle ventricular systolic dysfunction and calculating myocardial strain, exhibiting favorable consistency with echocardiography and other CMR techniques [15, 16, 17]. For atrial myocardial deformation, the fast long-axis method has shown superior stability, reliability, and reproducibility when compared to the FT method for obviating LA appendage and pulmonary veins [18, 19]. Moreover, LV global longitudinal strain (GLS), right ventricular (RV) GLS, and LA conduit strain showed significant prognostic value in individuals with DCM [20, 21, 22].
A comprehensive analysis of biatrial and biventricular myocardial deformation in DCM may contribute to improve risk stratification and implement treatment guidance. The aim of this study was to evaluate prognostic value in DCM patients by analyzing biventricular global radial strain (GRS), global circumferential strain (GCS), and GLS through FT, as well as biatrial GLS through fast long-axis method.
We retrospectively and consecutively enrolled participants who underwent CMR
from September 2020 to May 2022. Two cohorts were recruited: (i) patients with
non-ischemic DCM, which were defined as impaired systolic function with LVEF
CMR imaging was performed with a 3.0-Tesla scanner (Ingenia CX, Philips
Healthcare, the Netherlands) using a 32-channel phased-array abdomen coil. Cine
images were performed by using a steady-state free procession (SSFP) sequence
with multiple breath holds and electrocardiographic gating. The scanning
parameters were as follows: field of view (FOV) = 300
All CMR images were post-analyzed with a commercially available workstation (cvi42, Circle Cardiovascular Imaging Inc., Calgary, Alberta, Canada). The analysis of cardiac function and morphology was based on SSFP cine images, biventricular endocardial and epicardial borders, biatrial atrioventricular junctions, and midpoints of the posterior atrial wall. These measurements were tracked automatically based on manual calibration performed by two radiologists (each with 3 or 4 years of experience in CMR) who were blinded to baseline and outcome data. The produced measurements of left and right ventricular ejection fraction (RVEF), cardiac output (CO), end-diastolic volume (EDV), end-systolic volume (ESV), and atrial volumetric calculations. Atrial empty fraction (EF) was obtained according to the volumetric measurements. LV mass was calculated as myocardial gravity (1.05 g/mL) multiplied by total myocardium volume without papillary muscles.
CMR-FT derived strain parameters (radial strain, circumferential strain, and
longitudinal strain) were used to evaluate ventricular deformation capacity
through the use of cine images. Radial strain and circumferential strain were
measured via analysis of two-dimensional short-axis planes. Longitudinal strain
was measured by analyzing two-dimensional long-axis planes (2-chamber, 3-chamber,
and 4-chamber views) (Fig. 1). Atrial strain obtained by fast long-axis method
included longitudinal orientation. Atrial longitudinal strain was calculated
according to the distance between atrioventricular junction and midpoint of
posterior atrial wall (Fig. 1). The peak of the global curve was used as the
global strain value. LGE was quantified using mean
Representative images of ventricular and atrial strain measurement. (A) Feature tracking was used for the evaluation of biventricular GRS, GCS and GLS. Radial and circumferential strain were analyzed from short-axis planes, longitudinal strain was analyzed from long-axis planes. (B) Fast long-axis method was used for evaluating biatrial GLS according to the distance between atrioventricular junction and midpoint of posterior atrial wall. GCS, global circumferential strain; GLS, global longitudinal strain; GRS, global radial strain; LV, left ventricular; RV, right ventricular.
Clinical follow-up was performed via structured questionnaires [25] by telephone and then assessed by two experienced cardiologists. The primary endpoints included all-cause death, heart transplantation, and life-threatening arrhythmias. The secondary endpoint was hospitalization due to heart failure. Major adverse cardiac events (MACEs) were included in both primary and secondary endpoints.
Intra- and interobserver reproducibility for LV GLS were evaluated in 30 randomly selected study subjects. Intraobserver reproducibility was conducted 4 months later by a single radiologist who was blinded to the first analysis results. Interobserver reproducibility was assessed by two experienced radiologists who were blinded, without access to the other’s findings.
Normally distributed continuous variables were expressed as mean
The study consisted of 177 participants, including 124 DCM patients (70%), and
53 healthy volunteers (30%) (Fig. 2). In comparison to healthy volunteers, DCM
patients had higher prevalence of hypertension, diabetes mellitus,
hyperlipidemia, smoking, left bundle branch block
(LBBB), and presented more often with New York
Heart Association (NYHA) functional class
Inclusion criteria. This flow chart illustrates the inclusion and exclusion criteria used in the clinical study. Participants progressed through the study based on their eligibility and adherence to specific criteria. CMR, cardiovascular magnetic resonance; DCM, dilated cardiomyopathy; MACE, major adverse cardiac event.
Healthy (n = 53) | DCM (n = 124) | p* | No event (n = 104) | Event (n = 20) | p | ||
Age (years) | 53.17 |
52.82 |
0.873 | 52.66 |
53.65 |
0.75 | |
Male (n, %) | 28 (52.83) | 84 (67.74) | 0.059 | 69 (66.35) | 15 (75.00) | 0.448 | |
Height (cm) | 167.91 |
167.81 |
0.947 | 167.73 |
168.25 |
0.804 | |
Weight (kg) | 70.75 |
71.17 |
0.866 | 71.12 |
71.4 |
0.943 | |
BSA (m |
1.78 |
1.78 |
0.901 | 1.78 |
1.79 |
0.899 | |
HBP (n, %) | 3 (5.66) | 27 (21.77) | 0.009 | 24 (23.08) | 3 (15.00) | 0.561 | |
DM (n, %) | 2 (3.77) | 27 (21.77) | 0.003 | 21 (20.19) | 6 (30.00) | 0.377 | |
Hyperlipidemia (n, %) | 7 (13.20) | 91 (73.39) | 77 (74.04) | 14 (70.00) | 0.708 | ||
Smoking (n, %) | 6 (11.32) | 32 (25.80) | 0.032 | 29 (27.88) | 3 (15.00) | 0.228 | |
LBBB (n, %) | 0 (0) | 19 (15.32) | 0.003 | 19 (18.27) | 0 (0) | 0.041 | |
NYHA class | |||||||
I (n, %) | 53 (100) | 11 (8.87) | - | 11 (10.58) | 0 | - | |
II (n, %) | 0 | 24 (19.35) | - | 24 (23.08) | 0 | - | |
III (n, %) | 0 | 49 (39.52) | - | 39 (37.50) | 10 (50) | - | |
IV (n, %) | 0 | 40 (32.26) | - | 30 (24.19) | 10 (50) | - | |
NT-pro BNP (pg/mL) | 0 (0, 118) | 5802 (2176, 9179) | 1798 (887, 5145) | 7559 (3983, 12167) | |||
TnI (ug/L) | - | 0.02 (0, 0.07) | 0.02 (0, 0.06) | 0.04 (0, 0.56) | 0.07 | ||
CKMB (ug/L) | - | 0.80 (0.50, 1.50) | 0.80 (0.50, 1.40) | 1.10 (0.65, 1.68) | 0.262 | ||
ACEI/ARB (n, %) | 0 (0) | 104 (83.87) | 84 (80.77) | 20 (100.00) | 0.041 | ||
Beta blockers (n, %) | 0 (0) | 84 (67.74) | 73 (70.19) | 11 (55.00) | 0.183 | ||
Diuretics (n, %) | 0 (0) | 117 (94.35) | 97 (93.27) | 20 (100.00) | 0.232 | ||
Digoxin (n, %) | 0 (0) | 70 (56.45) | 58 (55.77) | 12 (60.00) | 0.727 |
p* indicates healthy versus DCM, p
The DCM cohorts were divided into event group and no event group, which included
20 subjects (16.13%) and 104 subjects (83.87%), respectively. There were no
significant differences in these two groups as related to age, sex, height,
weight, or incidences of hypertension, diabetes mellitus, hyperlipidemia, and
smoking (all p
In comparison to the healthy group, the DCM group exhibited several significant
differences. These included higher heart rate (HR) and left ventricular mass
index (LVMi), as well as lower CO, LVEF, RVEF, LA EF and right atrial (RA) EF
(all p
Healthy (n = 53) | DCM (n = 124) | p* | No event (n = 104) | Event (n = 20) | p | |
HR (1/min) | 69.00 |
83.00 |
80.00 |
96.00 |
0.005 | |
CO (L/min) | 6.04 |
5.35 |
0.028 | 5.37 |
5.25 |
0.814 |
LVMi (g/m |
47.93 |
79.32 |
78.67 |
82.72 |
0.395 | |
LVEF (%) | 66.93 |
21.98 |
23.43 |
14.45 |
||
LV EDVi (mL/m |
73.15 |
176.37 |
168.44 |
217.63 |
||
LV ESVi (mL/m |
24.72 |
142.84 |
133.36 |
192.13 |
||
LV EDD (mm) | 47.03 |
72.69 |
71.51 |
78.82 |
0.002 | |
RVEF (%) | 51.26 |
32.67 |
35.55 |
17.66 |
||
RV EDVi (mL/m |
62.99 |
95.17 |
89.85 |
122.84 |
||
RV ESVi (mL/m |
31.78 |
67.37 |
60.25 |
104.4 |
||
RV EDD (mm) | 27.6 |
27.49 |
0.939 | 26.55 |
32.4 |
0.027 |
LA EF (%) | 57.16 |
32.65 |
34.41 |
23.52 |
||
LAVi min (mL/m |
19.74 |
57.79 |
52.82 |
83.66 |
||
LAVi max (mL/m |
40.44 |
72.23 |
68.27 |
92.83 |
0.003 | |
RA EF (%) | 51 |
44.95 |
0.01 | 45.43 |
42.46 |
0.415 |
RAVi min (mL/m |
19.79 |
29.38 |
27.38 |
39.75 |
0.005 | |
RAVi max (mL/m |
33.37 |
38.83 |
0.013 | 37.01 |
48.34 |
0.071 |
LGE (%) | 0 (0, 0) | 6.70 (3.01, 12.87) | 6.13 (2.53, 12.61) | 10.36 (6.11, 14.86) | 0.056 | |
LV GRS (%) | 32.36 |
8.06 |
8.77 |
4.38 |
||
LV GCS (%) | –18.62 |
–6.18 |
–6.65 |
–3.72 |
||
LV GLS (%) | –17.57 |
–6.41 |
–6.98 |
–3.43 |
||
RV GRS (%) | 26.46 |
10.10 |
11.26 |
4.08 |
||
RV GCS (%) | –17.09 |
–5.19 |
–5.93 (–8.66, –3.75) | –1.79 (–3.77, 2.44) | ||
RV GLS (%) | –25.75 |
–14.05 |
–16.96 (–20.68, –12.95) | –10.00 (–14.66, 3.30) | ||
LA GLS (%) | 31.38 |
10.61 |
11.71 |
4.87 |
||
RA GLS (%) | 36.07 |
24.61 |
26.16 |
16.52 |
0.009 |
p* indicates healthy versus DCM, p
HR was higher in the event group compared to the no event group in the DCM
cohort (p
Biventricular GRS, GCS, and GLS were significantly impaired in the DCM group
compared to the healthy group (all p
Biventricular GRS was significantly lower in the event group relative to the no
event group (LV, 4.38
Comparison of biventricular GRS, GCS, GLS, biatrial GLS between no event group and event group. Values for the myocardial strain in the left (A) and right (B) side of the heart were statistically reduced in the event group relative to the no event group. GCS, global circumferential strain; GLS, global longitudinal strain; GRS, global radial strain; LA, left atrial; LV, left ventricular; RA, right atrial; RV, right ventricular.
Over a median follow-up period of 362 days
(IQR: 234 to 500 days), MACE occurred in 20 patients (16.13%) (all-cause death,
n = 5; heart transplantation, n = 1; life-threatening arrhythmia, n = 6;
hospitalizations due to heart failure, n = 8). The univariate Cox regression
analysis revealed that Ln (NT-pro BNP), LGE, HR, LV EDD, RV EDD, LV EF, LV EDVi,
LV ESVi, RV EF, RV EDVi, RV ESVi, LA EF, LAVi min, LAVi max, LV GRS, LV GCS, LV
GLS, RV GRS, RV GCS, RV GLS, LA GLS, RA GLS were all significant predictors of
MACE (all p
Univariate | Multivariate | |||
HR (95% CI) | p | HR (95% CI) | p | |
Ln (NT-pro BNP) | 1.907 (1.252, 2.904) | 0.003 | ||
LGE (%) | 1.051 (1.011, 1.094) | 0.013 | ||
Heart rate (1/min) | 1.021 (1.002, 1.041) | 0.033 | ||
LV EDD (mm) | 1.070 (1.030, 1.110) | |||
RV EDD (mm) | 1.054 (1.015, 1.096) | 0.007 | ||
LVEF (%) | 0.850 (0.774, 0.932) | 0.001 | 1.000 (0.873, 1.145) | 1.000 |
LV EDVi (g/m |
1.014 (1.007, 1.021) | |||
LV ESVi (g/m |
1.016 (1.009, 1.023) | |||
RVEF (%) | 0.928 (0.894, 0.964) | 0.948 (0.895, 1.004) | 0.067 | |
RV EDVi (g/m |
1.023 (1.011, 1.035) | |||
RV ESVi (g/m |
1.028 (1.016, 1.040) | |||
LA EF (%) | 0.930 (0.891, 0.972) | 0.001 | ||
LAVi min (g/m |
1.013 (1.008, 1.024) | |||
LAVi max (g/m |
1.014 (1.005, 1.022) | 0.002 | ||
LV GRS (%) | 0.616 (0.490, 0.774) | 0.060 (0.002, 2.387) | 0.134 | |
LV GCS (%) | 1.828 (1.370, 2.440) | 0.037 (0.000, 3.598) | 0.158 | |
LV GLS (%) | 1.854 (1.370, 2.510) | 1.788 (1.048, 3.049) | 0.033 | |
RV GRS (%) | 0.864 (0.792, 0.941) | 0.001 | 1.232 (0.952, 1.595) | 0.113 |
RV GCS (%) | 1.182 (1.070, 1.305) | 0.001 | 1.067 (0.815, 1.397) | 0.637 |
RV GLS (%) | 1.043 (1.017, 1.071) | 0.001 | 1.013 (0.974, 1.053) | 0.512 |
LA GLS (%) | 0.953 (0.912, 0.995) | 0.030 | ||
RA GLS (%) | 0.961 (0.928, 0.995) | 0.027 |
CI, confidence interval; EDD, end-diastolic diameter; EDVi, end-diastolic volume index; EF, empty fraction; ESVi, end-systolic volume index; GCS, global circumferential strain; GLS, global longitudinal strain; GRS, global radial strain; HR, hazard ratio; LA, left atrial; LAVi max, maximum left atrial volume index; LAVi min, minimum left atrial volume index; LGE, late gadolinium enhancement; LV, left ventricular; LVEF, left ventricular ejection fraction; MACEs, major adverse cardiac events; NT-pro BNP, N-terminal pro-B-type natriuretic peptide; RV, right ventricular; RVEF, right ventricular ejection fraction.
The ROC analysis revealed that the AUCs for the LA and RA GLS used to predict
MACEs were 0.762 and 0.701 respectively (both p
ROC showed significant prognostic values of biatrial GLS, biventricular GRS, GCS, GLS in predicting MACEs. The AUCs, sensitivity, specificity of the left (A) and right (B) side of the heart were displayed. GCS, global circumferential strain; GLS, global longitudinal strain; GRS, global radial strain; LA, left atrial; LV, left ventricular; MACE, major adverse cardiac event; RA, right atrial; ROC, receiver-operating characteristic; AUCs, area under the curves; RV, right ventricular.
Kaplan-Meier survival analyses for the prediction of MACEs. An
optimal cut-off value for LV GLS was identified at -4.81%, patients with a LV
GLS
In DCM patients, LV GLS showed strong negative correlations with LVEF and LA EF
(both p
LA GLS | LV GLS | |||
Parameter | r value | p value | r value | p value |
LV EDVi | –0.439 | 0.447 | ||
LV ESVi | –0.455 | 0.537 | ||
LAVi min | –0.720 | 0.508 | ||
LAVi max | –0.616 | 0.384 | ||
LVEF | 0.434 | –0.691 | ||
LA EF | 0.690 | –0.618 | ||
LVMi | –0.169 | 0.06 | 0.346 |
EDVi, end-diastolic volume index; EF, empty fraction; ESVi, end-systolic volume index; GLS, global longitudinal strain; LA, left atrial; LAVi max, maximum left atrial volume index; LAVi min, minimum left atrial volume index; LV, left ventricular; LVEF, left ventricular ejection fraction; LVMi, left ventricular mass index; DCM, dilated cardiomyopathy.
Measurement of cardiac strain parameters revealed solid reproducibility in healthy volunteers and DCM patients. The ICC for the intraobserver variability was found to be 0.933 (95% CI: 0.859–0.968) for LV GLS. Similarly, the ICC for the interobserver variability for LV GLS also had a high level of agreement with a value of 0.913 (95% CI: 0.827–0.958) (Fig. 6). These results indicate the reliability and consistency of the strain measurements in this patient population.
The high reproducibility of strain measurement in DCM and healthy groups. The intraobserver variability (A) and interobserver variability (B) of LV GLS in the two groups. DCM, dilated cardiomyopathy; GLS, global longitudinal strain; LV, left ventricular.
In this study, we employed FT and fast long-axis methods evaluate deformation characteristics of the ventricle and atrium in patients diagnosed with DCM. Our study yielded several important findings: The DCM group exhibited impaired biventricular GRS, GCS, GLS, and biatrial GLS in comparison to the healthy group. Furthermore, the event group demonstrated significantly reduced biventricular GRS, GCS, GLS, and biatrial GLS compared to the no event group. While the extent of LGE was higher in the DCM group compared to the healthy group, no statistical differences were observed between the event group and the no event group. Notably, LV GLS demonstrated significant and independent prognostic value outperforming other CMR parameters in predicting MACEs over a median follow-up of 362 days.
The reduction of biventricular GRS, GCS, GLS, and biatrial GLS were observed in the DCM group, which was consistent with a previous study about myocardial deformation characteristics [26]. In the early stage of DCM, cardiac function can be compensated by the Frank-Starling law, which enables an increase in myocardial contractility and helps to regulate reduced stroke volume [27]. However, in DCM patients who reached the end stage of disease, decreased cardiac function is consistently accompanied with enlargement of the ventricular chamber and decreased ventricular compliance [28].
The reduction in LV stroke volume contributes to an increase in preload, consequently leading to LA dysfunction [11, 29]. The interplay between LA and LV, pulmonary vascular pressure increases further contributes to RV dysfunction [9]. These mechanisms align with the findings of our study. The treatment methods for DCM include medication, cardioverter defibrillator implantation, resynchronization therapy, and heart transplantation. Early identification of disease characteristics and ultimate progression is crucial for guiding appropriate treatment.
LV GLS remains as an independent predicting parameter for other cardiac pathologies, including myocarditis, myocardial infarction, and even heart transplantation [30, 31, 32]. GLS refers to systolic shortening of the cardiac chamber in long-axis direction, which can be used to evaluate the motion ability of the ventricle in the cardiac cycle [33]. Raafs et al. [34] provided evidence that speckle tracking echocardiographic LV GLS emerged as an independent and incremental predictor of adverse outcome other than LVEF in patients with DCM. LV GLS has been suggested to be routinely measured for DCM prognosis assessment. Another study found RV GLS to be independently associated with MACEs independent of the of interaction between LA and RA [22], an event that is contradictory to our finding. We only found LV GLS to be an independent prognostic parameter that predicted MACE. There were also strong correlations between atrial and ventricular strain and CMR functional parameters, including LVEF and LA EF. Therefore, LA and RA-related parameters should not be underestimated, and all four chambers should be coordinated to assess overall pathological changes. The recognition of myocardial dysfunction in DCM patients is crucial for risk stratification and prediction of prognosis [35].
LGE was superior and independent from LVEF in predicting arrhythmic events, although LVEF was considered the main factor for selecting candidates for primary prevention with an implantable cardioverter-defibrillators [36]. Researchers have demonstrated that myocardial fibrosis is associated with increasing risk of ventricular arrhythmias [37]. The presence of LGE offers a powerful value in prognosis evaluation in non-ischemic cardiomyopathy [1, 2, 38]. In our study, LGE was only shown to be an important predictor of adverse outcomes under univariate Cox regression analysis. We believe this disparity might be explained by two points, (1) the relatively shorter follow-up period may account for the difference, (2) the total DCM cohort tended to be in the late stage of disease in our study population.
LA function is closely intertwined with LV function and impaired LV function is typically accompanied by a decrease in LA performance. The involvement of LA in regulating LV is divided into three phases: (1) reservoir phase, which involves the collection of pulmonary venous return during LV contraction; (2) conduit phase, where blood is passed to LV during early diastole; and (3) booster pump phase, which entails the augmentation of LV filling by atrial contraction during late diastole [39]. LA GLS has been extensively studied in various conditions including heart failure, atrial fibrillation, and myocardial fibrosis using multiple CMR techniques [40, 41]. Previous studies have examined LA reservoir strain, conduit strain, and booster strain to explore the influences of LA in different diseases [9, 18, 19]. In our study patients with DCM experienced significant impairments in both LA and LV function, making it challenging to differentiate LA function into these three distinct phases. Therefore, we solely considered LA GLS as a potential parameter that might influence the study results.
There were several limitations to this study. First, it was a single-center retrospective study including a small number of patients. Prospective research involving a larger study population is needed to validate our results. Second, by the study completion date, 8 parameters were included in the multivariate Cox regression analysis. For a more convincing statistical analysis, a larger study sample would be more appropriate. Third, the median follow-up period was 362 days (IQR: 234 to 500 days) and only 20 MACEs were recorded. A longer follow-up period should aid in the search for a prognosis response. Fourth, the majority of participants in our study were DCM patients with significantly reduced LVEF. It is important to note that our study specifically focused on DCM patients with severe systolic dysfunction, and therefore, the findings may not be directly applicable to those with mild to moderate impairment. Conducting a multi-center research study would be essential to mitigate this potential bias and provide more comprehensive insights. Finally, our study solely evaluated LGE through the quantification of its volumetric proportion of the total LV myocardium. The pattern or distribution of LGE might offer additional novel viewpoints. Multiple LGE-related parameters have the potential to provide insights leading to improved clinical outcomes.
By using FT and fast long-axis method derived from CMR, we found that biventricular GRS, GCS, GLS, and biatrial GLS were significantly impaired in the event group relative to the no event group in DCM. LV GLS was independently associated with MACE in DCM patients. Comprehensive CMR examination should be systematically performed, in order to understand disease characteristics, as well as improve the risk stratification and therapeutic management for patients with DCM.
The datasets supporting the conclusions of the current study are available from the corresponding author on reasonable request.
SLL and YLL designed the study and wrote the paper. JXL, XYW, YL, JYR and YMZ post-processed the images and drafted the work. XG, JJC, QL, DW and BX collected the data and revised the manuscript. XDW and ZYW analyzed the data and drafted the work. GKW and BY designed the study and revised the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.
The study was performed according to the Declaration of Helsinki and was approved by the Committee of the Second Affiliated Hospital of Harbin Medical University (KY2021-132). All participants gave written informed consent.
The authors appreciate Linlin Dai for the assistance in diagnostic guidance.
This study was supported by the National Natural Science Foundation of China (Grant No. 82000330/XDW and Grant No. 82100529/XG).
The authors declare no conflict of interest.
Publisher’s Note: IMR Press stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.