Assessment of left atrial native T1 values using non-contrast 3.0 T cardiac magnetic resonance imaging: a feasibility and reproducibility study

Abstract

Purpose

Atrial cardiomyopathy (ATC) is characterized by structural and functional atrial remodeling and is associated with adverse cardiovascular outcomes. Native T1 mapping has emerged as a sensitive technique for tissue characterization; however, data on left atrial (LA) native T1 mapping remain limited. We evaluated the feasibility and reproducibility of LA T1 measurements using non-contrast 3.0-T cardiac magnetic resonance (CMR), thereby providing a methodological foundation for future studies of atrial tissue characterization.

Methods

Healthy volunteers in sinus rhythm underwent non-contrast 3.0-T CMR, including native T1 mapping using a modified Look-Locker inversion recovery (5s(3s)3s MOLLI) sequence. LA native T1 values were measured in the 2-, 3-, and 4-chamber long-axis views. Two independent specialists measured LA native T1 values using regions of interest placed within the inner 2 reconstructed pixels of the LA wall to minimize in-plane partial volume effects. Feasibility was assessed by the proportion of measurable studies, and reproducibility by intra- and inter-observer intraclass correlation coefficients (ICC). Additional analyses were performed to assess potential blood-pool contamination.

Results

Forty volunteers (50% female; mean age 33.6 ± 6.1 years; body mass index of 22.7 ± 3.2 kg/m²) were included. LA native T1 values were measurable in 62.5%, 95.0%, and 75.0% of subjects in the 2-, 3-, and 4-chamber views, respectively, identifying the 3-chamber view as the most feasible imaging plane. Mean LA native T1 values were consistent across views (1,441–1,448 ms; all p > 0.05) without gender difference. Reproducibility was excellent, with intra-observer ICCs of 0.921–0.940 and inter-observer ICCs of 0.894–0.920. Agreement between views was good (ICC 0.709–0.878) without significant proportional bias. Blood-pool T1 values were substantially higher than LA T1 values (1,859–1,868 ms vs. 1,441–1,448 ms) and showed no significant correlation. Validation analyses demonstrated a progressive increase in blood-pool T1 values with increasing distance from the LA wall, supporting measurement specificity.

Conclusion

Non-contrast T1 mapping enables feasible and highly reproducible quantification of LA native T1 values. The 3-chamber view demonstrated the highest measurability and may be the preferred imaging plane for LA tissue characterization. These findings provide a methodological framework for future investigations of ATC and other atrial diseases.

Clinical trial number

Not applicable.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12880-026-02635-2.

Introduction

The left atrium (LA) plays a crucial role in cardiovascular physiology, functioning as a reservoir, conduit, booster pump, and integral component of the cardiac conduction system. Atrial cardiomyopathy (ATC) involves structural, functional, and electrophysiological alterations of the atrium that may lead to clinically significant manifestations and adverse cardiovascular outcomes [1]. Various cardiac and systemic conditions, including atrial fibrillation (AF), heart failure, ventricular cardiomyopathies, valvular heart diseases, cardiac amyloidosis, aging, hypertension, and diabetes, have been implicated in the development of ATC [1–3]. Early identification and treatment for ATC may provide opportunities to improve patient prognosis.

Several imaging modalities and parameters have been used to evaluate ATC. Echocardiography remains the most widely used technique, providing assessment of LA size, volume, strain, and left ventricular (LV) diastolic function through Doppler-based measurements [4]. However, direct tissue characterization of the atrial myocardium remains challenging because the LA wall is very thin, with a reported mean thickness of 1.6–2.8 mm on computed tomography [5]. Cardiac magnetic resonance imaging (CMR) has emerged as a promising imaging modality for atrial assessment. Beyond chamber quantification, late gadolinium enhancement (LGE) CMR provides incremental prognostic information and clinical relevance [6, 7]. The presence of LA LGE is associated with sinus node dysfunction, stroke, and progression of AF [8–10]. Nevertheless, atrial fibrosis detected by LGE and structural chamber abnormalities generally represent relatively advanced stages of ATC [11]. Consequently, more sensitive imaging techniques are needed to detect earlier atrial myocardial remodeling.

Native T1 mapping has become pivotal in quantitatively assessing interstitial fibrosis by measuring myocardial longitudinal relaxation times. This technique offers superior sensitivity compared to the traditional LGE, allowing for early detection of subtle changes in extracellular matrix composition associated with fibrosis in various cardiac conditions, particularly the diseases involving LV [12]. However, data regarding native T1 mapping of the LA in humans remain limited. Furthermore, a standardized methodology for measuring native T1 values within the thin atrial wall has not been established. Therefore, we sought to evaluate and validate a feasible and reproducible methodology for measuring LA native T1 values in healthy volunteers using 3.0-T CMR T1 mapping, thereby providing a methodological foundation for future studies of LA tissue characterization in patients with ATC and other atrial diseases.

Methods

Study population

Forty healthy volunteers (20 females and 20 males) were prospectively recruited to undergo 3.0-T CMR at King Chulalongkorn Memorial Hospital. The inclusion criteria were: (i) age 18 years or older, (ii) BMI range from 18.5 to 29.9 kg/m², (iii) no known underlying disease and cardiovascular risk factors, and (iv) normal blood tests in the previous year. The exclusion criteria were: (i) pregnancy, (ii) claustrophobia, (iii) metallic implantation, (iv) AF or other rhythm abnormalities identified during CMR acquisition, and (v) structural abnormality on the CMR findings. Age, gender, body mass index (BMI), and CMR parameters were collected. The study protocol was approved by the Institutional Review Board, Faculty of Medicine, Chulalongkorn University. Informed consent was obtained for study in human subjects. All procedures performed in studies involving human participants were in accordance with the World Medical Association’s Declaration of Helsinki.

CMR protocol

The scan was performed using 3 Tesla (Magnetom Vida; Siemens Healthineers, Erlangen, Germany) with an 18-channel cardiac phased array receiver. The protocol consisted of a localizer, bright-blood axial images covering the thoracic inlet to the upper abdomen, cine images, and native T1 mapping. A retrospective ECG-gated, end-expiratory breath-hold cine balanced steady-state free precession (bSSFP) accelerated using compressed sensing with an acceleration factor of 9.5 was performed to acquire long-axis (2-, 3-, and 4-chamber) cine images with a slice thickness of 6 mm, as well as contiguous short-axis cine images covering the whole heart with an 8-mm slice thickness and no gaps (Supplementary Table 1). Native T1 values of the LA were measured using non-contrast T1 mapping acquired in the 2-, 3-, and 4-chamber long-axis views with a modified Look-Locker inversion recovery (5s(3s)3s MOLLI) sequence [13, 14]. T1 maps were reconstructed using inline motion correction and included T1 error maps, T1* maps and T1 maps. Typical acquisition parameters consisted of a single-shot bSSFP readout with a trigger delay timed to mid-to-end diastole. For RR intervals > 700 ms, imaging parameters were: flip angle 35°, bandwidth 1085 Hz/Px, slice thickness 6 mm, acquisition matrix 144 × 256, reconstruction matrix 218 × 256, acquisition voxel size 2.1 × 1.4 × 6.0 mm³, reconstructed voxel size 1.4 × 1.4 × 6.0 mm³_, TE 1.12 ms, TR 280.6 ms, and field of view (FOV) 306.6 × 360 mm². For RR interval ≤ 700 ms, imaging parameters were: flip angle 35°, bandwidth 1085 Hz/Px, slice thickness 6 mm, acquisition matrix 132 × 192, reconstruction matrix 164 × 192, acquisition voxel size 2.3 × 1.9 × 6.0 mm³, reconstructed voxel 1.9 × 1.9 × 6.0 mm3, TE 1.0 ms, TR 263 ms, and FOV 307.5 × 360 mm².

Post-processing CMR analysis

All images were analyzed using dedicated post-processing software (syngo.via, Siemens Healthineers, Erlangen, Germany). The LA endocardial border (excluding the LA appendage and pulmonary veins), the LV endocardial border (including papillary muscles), and LV epicardial border were manually delineated on the entire stack of short-axis cine images at both end-diastole and end-systole to quantify chamber volumes and ejection fractions [15, 16]. The LA anteroposterior diameter was measured at the level of the aortic sinus of Valsalva on the 3-chamber long-axis view. All chamber volumes were indexed to body surface area.

Before native T1 analysis, all source images from the T1 mapping sequence were carefully reviewed, and images with visible misregistration artifacts were excluded. Two independent CMR specialists measured LA native T1 values by manually placing regions of interest (ROIs) within the LA myocardium on the 2-, 3-, and 4-chamber grayscale T1 maps. To assess reproducibity across image display formats, the same analysis was repeated using color-encoded T1 maps.

Each image was magnified to optimize visualization of the LA myocardium. A single ROI was drawn along the entire visible length of the LA wall, with a minimum ROI size of 20 reconstructed pixels [17]. To minimize measurement error, ROIs were not placed within the interatrial septum because of its relatively high interstitial fat content [18]. More importantly, given the spatial resolution of the acquired images, ROIs were restricted to the inner two reconstructed pixels of the LA wall to maximize separation from tissue interfaces susceptible to in-plane partial volume averaging from the adjacent blood pool and surrounding extra-myocardial tissues [19]. (Fig. 1). LA segments were considered unmeasurable if the ROI size was less than 20 reconstructed pixels on T1 maps.

Fig. 1.

Measurements of left atrial native T1 values from different long-axis views

LA, left atrium; LV, left ventricle

In addition, LA wall thickness was directly measured on the corresponding long-axis cine bSSFP images to assess the potential influence of in-plane partial volume effect on LA T1 measurements (Supplementary Fig. 1A). To further evaluate potential blood-pool contamination, a validation analysis was performed by placing serial ROIs within the adjacent blood pool, progressing incrementally from the LA wall toward the center of the LA cavity on each long-axis grayscale T1 map. A separate ROI was also placed in the mid-LA cavity to determine native blood-pool T1 values (Supplementary Fig. 1B).

Statistical analysis

Continuous data were presented as mean with standard deviation (SD). The Shapiro-Wilk test was used as a prerequisite check to assess the assumption of normality before applying parametric statistical methods. An independent t-test was used to compare LA native T1 values between genders and between long-axis views. The feasibility of each view (2-, 3-, and 4-chamber) was determined by the number of cases in which the native T1 values could be adequately measured. Intra- and inter-observer agreements using an intraclass correlation coefficient (ICC) with a two-way random-effects model were analysed to determine the reproducibility. Intra-observer agreement was derived from the repeated analysis after four weeks in all subjects, while two independent CMR specialists analysed 20 randomly selected subjects to assess inter-observer variability. A mean ± 2SD of LA native T1 values was used to define the reference range. ICC and Bland-Altman plots were used to test agreements between LA native T1 values from different long-axis views. Statistical significance was indicated by a p-value of less than 0.05. All analyses were processed with SPSS statistics software version 25.0 (IBM Corporation, Armonk, New York).

Results

Forty-two volunteers were screened for eligibility. Two volunteers were excluded (one due to an exceeding BMI and one with an incidental atrial septal defect detected from CMR), leaving 40 volunteers (20 males and 20 females) enrolled in the study. The mean age and BMI were 33.6 ± 6.1 years and 22.7 ± 3.2 kg/m², respectively. The conventional LA and LV parameters, including volume, mass, and function, are shown in Table 1. Briefly, the mean LV ejection fraction and LV stroke volume were 55.8 ± 3.2% and 77 ± 15.7 ml, respectively. The mean LV native T1 was 1,237 ± 40 ms (normal range 1228 ± 48 ms for our center [14]). The mean LA diameter, LA maximal volume index, and LA ejection fraction were 28 ± 4.4 mm, 35 ± 7.4 ml/m², and 48 ± 5.0%, respectively.

Table 1.

Baseline characteristics

	Total (n = 40)	Female (n = 20)	Male (n = 20)
Demographic
Age (year)	33.6 ± 6.1	33.5 ± 6.6	33.8 ± 5.8
BMI (kg/m2)	22.7 ± 3.2	21.6 ± 3.2	23.7 ± 2.9
BSA (m2)	1.7 ± 0.2	1.6 ± 0.1	1.8 ± 0.1
LV parameters
LVEF (%)	55.8 ± 3.2	55.8 ± 3.4	55.7 ± 3.0
LVEDV (ml)	138 ± 27.4	120 ± 20.5	156 ± 20.3
LVEDVi (ml/m2)	81 ± 12.0	75 ± 10.0	86 ± 11.5
LVESV (ml)	61 ± 13.1	53 ± 10.8	69 ± 9.9
LVESVi (ml/m2)	36 ± 5.7	33 ± 5.2	38 ± 5.3
LVSV (ml)	77 ± 15.7	67 ± 11.1	87 ± 12.5
LVSVi (ml/m2)	45 ± 7.3	42 ± 6.1	48 ± 7.3
LVCO (l/min)	5.1 ± 0.9	4.7 ± 0.8	5.5 ± 0.8
LVCOi (l/min/m2)	3.0 ± 0.5	3.0 ± 0.5	3.0 ± 0.5
LVMI (g/m2)	46 ± 8.7	41 ± 4.3	52 ± 8.1
LV myocardium native T1 (ms)	1,237 ± 40	1,254 ± 40	1,220 ± 32
LA parameters
LA diameter (mm)	28 ± 4.4	28 ± 3.4	28 ± 5.4
LAEF (%)	48 ± 5.0	51 ± 4.6	46 ± 4.3
LA maximal volume (ml)	60 ± 14.8	53 ± 9.0	66 ± 16.7
LA maximal volume index (ml/m2)	35 ± 7.4	33 ± 5.6	36 ± 8.7
LA minimal volume (ml)	31 ± 9.5	26 ± 5.6	36 ± 10.4
LA minimal volume index (ml/m2)	18 ± 4.9	17 ± 3.7	20 ± 5.5

Continuous data are presented as mean ± SD

BMI, body mass index; BSA, body surface area; LA, left atrium; LAEF, left atrium ejection fraction; LV, left ventricle; LVCO, left ventricular cardiac output; LVCOi, left ventricular cardiac output index; LVEDV, left ventricular end-diastolic volume; LVEDVi left ventricular end-diastolic volume index; LVEF, left ventricular ejection fraction; LVESV, left ventricular end-systolic volume; LVESVi left ventricular end-systolic volume index; LVMI, left ventricular mass index; LVSV, left ventricular stroke volume; LVSVi, left ventricular stroke volume index

Feasibility of acquiring LA native T1 values

Table 2A shows that LA native T1 values were measurable in 25 (62.5%), 38 (95%), and 30 (75%) volunteers in the 2-, 3-, and 4-chamber views, respectively, with consistent measurability across both grayscale and color-encoded images. In one volunteer (2.5%), LA native T1 values could not be measured in any long-axis view due to the LA wall was too thin, resulting in an ROI of fewer than 20 reconstructed pixels for T1 analysis. The mean LA native T1 values derived from grayscale images in the 2-, 3-, and 4-chamber views were 1,447 ± 33, 1,441 ± 33, and 1,448 ± 35 ms, respectively. Corresponding values obtained from color-encoded images were 1,448 ± 32, 1,445 ± 33, and 1,450 ± 32 ms, respectively. No significant differences in LA native T1 values were observed between females and males or among the three long-axis views (p-value = 0.458 for 2-chamber vs. 3-chamber, 0.780 for 2-chamber vs. 4-chamber, and 0.360 for 3-chamber vs. 4-chamber).

Table 2.

Native T1 values of LA wall and blood in mid LA chamber

A. Native T1 values of LA wall
	Measurable /total subject	Total	Female	Male	p-value*	Intra-observer agreement (95% CI)	Inter-observer agreement (95% CI)
Analysis based on grayscale image
LA native T1, 2-chamber (ms)	25/40 (62.5%)	1,447 ± 33	1,449 ± 29	1,446 ± 36	0.856	0.921 (0.821, 0.965)	0.917 (0.671, 0.979)
LA native T1, 3-chamber (ms)	38/40 (95%)	1,441 ± 33	1,445 ± 27	1,437 ± 39	0.463	0.928 (0.858, 0.963)	0.894 (0.588, 0.973)
LA native T1, 4-chamber (ms)	30/40 (75%)	1,448 ± 35	1,457 ± 34	1,440 ± 35	0.197	0.937 (0.866, 0.970)	0.904 (0.606, 0.976)
Analysis based on color-encoded image
LA native T1, 2-chamber (ms)	25/40 (62.5%)	1,448 ± 32	1,449 ± 35	1,448 ± 31	0.949	0.931 (0.844, 0.969)	0.916 (0.775, 0.969)
LA native T1, 3-chamber (ms)	38/40 (95%)	1,445 ± 33	1,451 ± 26	1,439 ± 37	0.247	0.940 (0.885, 0.969)	0.920 (0.786, 0.970)
LA native T1, 4-chamber (ms)	30/40 (75%)	1,450 ± 32	1,456 ± 32	1,446 ± 32	0.423	0.928 (0.850, 0.966)	0.901 (0.526, 0.970)

B. Native T1 values of blood in mid-LA chamber
		Total (n = 40)	Female (n = 20)	Male (n = 20)	p-value*	Intra-observer agreement (95% CI)	Inter-observer agreement (95% CI)
Analysis based on grayscale image
Blood native T1, 2-chamber (ms)		1,868 ± 68	1,900 ± 72	1,837 ± 47	0.002	0.997 (0.991, 0.999)	0.985 (0.941, 0.996)
Blood native T1, 3-chamber (ms)		1,868 ± 66	1,889 ± 71	1,845 ± 53	0.034	0.997 (0.992, 0.999)	0.949 (0.793, 0.987)
Blood native T1, 4-chamber (ms)		1,859 ± 66	1,886 ± 71	1,833 ± 49	0.006	0.998 (0.994, 0.999)	0.989 (0.959, 0.997)
Analysis based on color-encoded image
Blood native T1, 2-chamber (ms)		1,868 ± 69	1,899 ± 73	1,837 ± 47	0.003	0.997 (0.993, 0.999)	0.995 (0.980, 0.999)
Blood native T1, 3-chamber (ms)		1,867 ± 66	1,889 ± 72	1,846 ± 54	0.033	0.996 (0.991, 0.999)	0.986 (0.945, 0.996)
Blood native T1, 4-chamber (ms)		1,859 ± 66	1,886 ± 70	1,831 ± 50	0.009	0.996 (0.989, 0.998)	0.982 (0.932, 0.995)

Continuous data are presented as mean ± SD

CI, confident interval; LA, left atrium

* Independent t-test between female and male

The direct measurements of LA wall thickness on the corresponding long-axis cine bSSFP images showed mean thickness of 2.42 ± 0.29, 2.78 ± 0.31, and 2.39 ± 0.27 mm in the 2-, 3-, and 4-chamber views, respectively (Supplementary Tables 2 and Supplementary Fig. 1A).

A subgroup analysis was performed according to heart rate during image acquisition (RR interval > 700 ms vs. ≤700 ms). The mean heart rate of the study cohort was 78 ± 11 bpm (range 60–97 bpm). LA native T1 values were generally comparable between the two groups. No significant differences were observed in the 2-chamber view (1,453 ± 33 ms vs. 1,432 ± 27 ms; p = 0.198), the 3-chamber view (1,447 ± 36 ms vs. 1,451 ± 21 ms; p = 0.106), or the 4-chamber view (1,448 ± 26 ms vs. 1,458 ± 26 ms; p = 0.439) (Supplementary Table 3).

Intra- and inter-observer variability

The ICC for intra- and inter-observer agreements for analysis on grayscale images were 0.921 and 0.917 for the 2-chamber view, 0.928 and 0.894 for the 3-chamber view, and 0.937 and 0.904 for the 4-chamber view. The ICC for intra- and inter-observer agreements for analysis on color-encoded images were 0.931 and 0.916 for the 2-chamber view, 0.940 and 0.920 for the 3-chamber view, and 0.928 and 0.901 for the 4-chamber view (Table 2).

The ICC of LA native T1 values measured on grayscale images from the 2-chamber vs. 3-chamber, 2-chamber vs. 4-chamber, and 3-chamber vs. 4-chamber views were 0.772 (95% CI: 0.484, 0.900), 0.809 (95% CI: 0.496, 0.927), and 0.818 (95% CI: 0.617, 0.914), respectively. The ICC of T1 values measured on color-encoded images from the 2-chamber vs. 3-chamber, 2-chamber vs. 4-chamber, and 3-chamber vs. 4-chamber views were 0.709 (95% CI: 0.329, 0.874), 0.878 (95% CI: 0.680, 0.953), and 0.834 (95% CI: 0.649, 0.922), respectively.

Bland-Altman plots demonstrated an acceptable agreement in all comparison pairs without significant proportional bias (Fig. 2).

Fig. 2.

Bland-Altman plots depict agreements between left atrial native T1 values measured from different long-axis views. Upper row: analysis performed on grayscale images; lower row: analysis performed on color-encoded images

Validation analysis of potential blood-pool contamination

Table 2B demonstrates that the mean native T1 values of blood in the mid-LA chamber on grayscale images were 1,868 ± 68 ms in the 2-chamber view, 1,868 ± 66 ms in the 3-chamber view, and 1,859 ± 66 ms in the 4-chamber view. Blood-pool T1 values were significantly higher in females than in males across all long-axis views (p = 0.002 for 2-chamber, p = 0.034 for 3-chamber, and p = 0.006 for 4-chamber). Supplementary Fig. 2 shows no significant correlation between LA myocardial and blood-pool T1 values in any long-axis view (r = 0.287; p = 0.124 for 4-chamber, r = 0.253; p = 0.125 for 3-chamber, and r = 0.316; p = 0.124 for 2-chamber). Serial ROIs placed within the blood pool adjacent to the LA wall demonstrated a progressive increase in T1 values with increasing distance from the LA wall toward the center of the LA cavity, a pattern consistently observed across all long-axis views (Supplementary Fig. 1B and Supplementary Fig. 3). Notably, ROIs positioned at the LA–blood interface, which were excluded from myocardial analysis, exhibited intermediate T1 values, consistent with signal averaging between the atrial wall and adjacent blood pool.

Discussion

Our findings demonstrate that LA native T1 mapping is a feasible and reproducible technique for quantifying native T1 values in healthy individuals. To our knowledge, this is the first study to systematically evaluate and validate the feasibility and techniques used for measuring LA native T1 value in humans.

Feasibility

LA native T1 values were successfully measured in nearly all participants, with only one subject deemed unmeasurable in all long-axis views because of an extremely thin atrial wall. LA segments were considered measurable only when the ROI contained at least 20 reconstructed pixels, an objective criterion that minimizes measurement error associated with very small ROIs and is consistent with current consensus recommendations for T1 mapping [17]. Among the three acquisition planes, the 3-chamber view demonstrated the highest measurability, followed by the 4- and 2-chamber views. Mean LA native T1 values were consistent across all views, with no significant differences in either the mean values or measurement variability. Similarly, excellent agreement was observed between measurements obtained from grayscale and color-encoded T1 maps. These findings suggest that LA native T1 quantification is robust across different long-axis views and image display formats when standardized acquisition and analysis methods are applied. Given its superior measurability, the single 3-chamber view may represent the preferred acquisition plane for clinical implementation, particularly when scan time is limited.

LGE-CMR can identify atrial replacement fibrosis and provides incremental predictive value beyond conventional chamber quantification [1, 9]. However, the technique requires accurate selection of inversion time and reliable identification of normal myocardium for signal nulling [20]. More importantly, abnormalities detected by LGE-CMR generally reflect relatively advanced stages of ATC [11]. Consequently, parametric T1 mapping has emerged as a more sensitive technique for detecting early extracellular matrix remodeling [21] and has become an established tool for diagnosis and risk stratification of a variety of LV diseases [22–26]. In contrast, T1 mapping of the LA remains substantially less explored.

To date, only two studied have investigated the clinical significance of post-contrast LA T1 mapping [27, 28]. Beinart et al. demonstrated the feasibility of post-contrast LA T1 measurements using 1.5-T CMR and reported a strong association with intracardiac electrogram voltage amplitude [27]. Ling et al. examined post-contrast LA T1 values in relation to clinical outcomes after AF ablation and found that shorter T1 times associated with longer AF duration, while values above 230 ms predicted freedom from AF [28]. Unlike our approach, both studies measured T1 values at the interatrial septum on the 4-chamber view. This methodology may be more susceptible to measurement variability due to interstitial fat infiltration, particularly in patients with lipomatous hypertrophy of the interatrial septum [18].

Despite the potential clinical value of contrast-enhanced LA T1 mapping, native T1 mapping offers several practical advantages, particularly in patients with contraindications to gadolinium administration or those requiring serial follow-up examination. To date, only two studied have investigated LA native T1 mapping [29, 30]. Bouazizi et al. provided histological evidence in an ex vivo CMR study, showing strong correlations between LA native T1 values and histological tissue composition, including interstitium, fatty fibrosis, adipose tissue, and overall components (all r > 0.90) [29]. Luetkens et al. reported significantly higher LA native T1 values in patients with recurrent AF one year after catheter ablation and demonstrated that LA native T1 independently predicted AF recurrence (hazard ratio 1.007, p = 0.001) [30]. Collectively, these findings suggest that LA native T1 values may serve as a non-invasive marker of atrial remodelling, with potential applications in the early detection, phenotyping, and risk stratification of patients with ATC and other atrial disorders. However, while initial findings are promising, larger prospective studies are needed to establish its clinical utility and prognostic value.

We observed substantially higher native T1 values in the LA than those typically reported for the LV at 3.0 T (1,441–1,448 ms vs. approximately 1,237 ms). Several biological differences between atrial and ventricular myocardium may account for this observation. Atrial cardiomyocytes are smaller than ventricular cardiomyocytes [1], and the extracellular matrix composition of the atrial wall differs considerably from that of the ventricle. A histopathology-based study has shown substantially greater collagen content in the atrial myocardium, with more than twice the collagen fraction observed in ventricular tissue [31]. Consistent with these findings, previous post-contrast LA T1 study reported lower post-contrast T1 values in the LA than in the LV, suggesting greater gadolinium accumulation and a larger extracellular matrix volume within the atrial wall [27]. Additionally, we observed no significant sex-related differences in LA native T1 values, in contrast to LV native T1 values, which are typically higher in females than in males [14].

Reproducibility

The LA native T1 measurement approach used in this study demonstrated good intra- and inter-observer reproducibility. Several methodological features likely contributed to this performance, including: (i) image magnification to optimize visualization of the thin atrial wall, (ii) placement of ROIs along the entire visible length of the LA wall to avoid very small sampling regions, (iii) maintenance of adequate separation from tissue interfaces prone to partial volume effects, and (iv) avoidance of the interatrial septum to minimize contamination from adipose tissue. Standardization of ROI placement and image interpretation likely contributed to the excellent reproducibility observed in this study and may facilitate broader adoption of LA T1 mapping in future multicenter investigations. Similar findings were reported by Beinart et al., who demonstrated intra- and inter-observer ICCs exceeding 0.90 for post-contrast LA T1 measurements [27]. Collectively, these findings indicate that contemporary CMR image quality is sufficient to enable reliable quantification of thin atrial structures.

Validation analysis to differentiate the LA wall from the blood pool

Because the in-plane spatial resolution of the T1 maps was lower than that of the corresponding cine bSSFP images, the LA–blood interface was less sharply defined and the apparent LA wall appeared thicker on the T1 maps than on the cine images (Supplementary Fig. 1). Direct measurements on cine images demonstrated a mean LA wall thickness of only 2.4–2.8 mm, highlighting the inherent susceptibility of LA T1 measurements to partial volume effects from the adjacent blood pool. To mitigate in-plane partial volume effects, we restricted ROI placement to the inner two reconstructed pixels of the visible LA wall, such that the inner ROI boundary closely approximated the anatomical LA wall visualized on the corresponding cine image (Supplementary Fig. 1A). We then performed several validation analyses to determine whether the measured LA native T1 values predominantly reflected atrial myocardial tissue rather than blood-pool signal contamination.

First, LA myocardial T1 values (1,441 ± 33 to 1,448 ± 35 ms) were substantially lower than blood-pool T1 values (1,859 ± 66 to 1,868 ± 68 ms), demonstrating clear separation between the two tissues. Furthermore, no significant correlation was observed between LA myocardial and blood-pool T1 values across any long-axis view. Blood-pool T1 values were higher in females, consistent with previous reports and likely reflecting sex-related differences in hematocrit and blood iron composition [14]. In contrast, LA myocardial T1 values did not differ significantly between sexes. This discordance further supports the finding that the measured LA T1 values were not primarily driven by blood-pool contamination, as substantial contamination would be expected to transmit these sex-related differences to the myocardial measurements.

Second, serial ROIs placed progressively within the blood pool adjacent to the LA wall demonstrated a gradual increase in T1 values with increasing distance from the atrial wall toward the center of the LA cavity, a pattern that was consistently observed across all long-axis views. Furthermore, ROIs positioned at the LA–blood interface, which were deliberately excluded from myocardial analysis, yielded intermediate T1 values, as expected from partial volume averaging of atrial myocardial and blood-pool signals. Collectively, these findings support that the measured LA native T1 values represent a reproducible surrogate of atrial myocardial tissue characteristics rather than being primarily influenced by blood-pool signal contamination.

Limitations

This study has several limitations. First, the healthy volunteer cohort was relatively young and had a normal body mass index, which may limit the generalizability of our findings to older populations. Age-related increases in atrial adipose tissue, particularly within the interatrial septum, may influence LA native T1 values [18]. Second, as with LV native T1 mapping, LA T1 values are expected to vary among centers because differences in magnetic field strength, scanner vendors, acquisition parameters, and pulse sequences [17, 32]. Therefore, the values reported in this study should not be considered universal reference ranges. Third, LA native T1 map was acquired during ventricular diastole, the standard acquisition phase for LV myocardial T1 assessment. However, the optimal cardiac phase for LA wall evaluation remains uncertain because atrial mechanics differ from ventricular contraction patterns and LA wall thickness varies throughout the cardiac cycle. Consequently, ventricular diastole may not necessarily represent the optimal phase for LA tissue characterization. Future studies should investigate whether alternative acquisition windows can improve LA wall visualization, increase measurable wall thickness, and further enhance the reproducibility of LA T1 measurements. Forth, although we implemented an ROI placement strategy designed to minimize partial volume effects, neither in-plane nor through-plane partial volume averaging associated with the 6-mm slice thickness could be completely eliminated. Furthermore, the residual degree of in-plane partial volume effect could not be reliably quantified because vendor-generated parametric T1 maps do not preserve the spatial frequency information required to estimate the effective point spread function (PSF) using full width at half maximum (FWHM) analysis. In addition, the inline reconstruction process, including motion correction and T1 curve fitting, further alters the effective spatial resolution. Therefore, the reported LA native T1 values should be interpreted as a clinically feasible and reproducible surrogate of atrial myocardial tissue characteristics rather than a direct histological equivalent of pure atrial myocardium. Future advances, including thinner slice acquisition and higher-resolution T1 mapping techniques, may further improve measurement accuracy. Finally, the study population consisted exclusively of healthy individuals. Therefore, the clinical applicability, diagnostic performance, and prognostic value of LA native T1 values in patients with ATC and other atrial diseases remain to be established.

Conclusion

Non-contrast T1 mapping enables feasible and highly reproducible quantification of LA native T1 values. The 3-chamber view demonstrated the highest measurability and may represent the preferred imaging plane for LA tissue characterization. These findings provide a methodological framework for future investigations of ATC and other atrial diseases. Further studies in larger and more diverse patient populations are warranted to determine the clinical utility and prognostic significance of LA native T1 values.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ATC

Atrial cardiomyopathy

AF

Atrial fibrillation

BMI

Body mass index

CMR

Cardiac magnetic resonance imaging

ICC

Intraclass correlation coefficient

LA

Left atrium

LAEF

Left atrium ejection fraction

LGE

Late gadolinium enhancement

LV

Left ventricle

LVCO

Left ventricular cardiac output

LVCOi

Left ventricular cardiac output index

LVEDV

Left ventricular end-diastolic volume

LVEDVi

Left ventricular end-diastolic volume index

LVEF

Left ventricular ejection fraction

LVESV

Left ventricular end-systolic volume

LVESVi

Left ventricular end-systolic volume index

LVMI

Left ventricular mass index

LVSV

Left ventricular stroke volume

LVSVi

Left ventricular stroke volume index

ROI

Region of interest

SD

Standard deviation

Author contributions

NT, DW and PC contributed to the study conception and design. Material preparation, data collection and analysis were performed by NT, NN, PS, ST, RP, YV and MT. The first draft of the manuscript was written by NT and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Data availability

The datasets used and/or analyzed during this study are available from the corresponding author upon reasonable request.

Declarations

Ethical approval and consent to participate

Ethical approval by the Institutional Review Board, Faculty of Medicine, Chulalongkorn University were obtained prior to data collection and analysis. Informed consent was obtained for study in human subjects. All procedures performed in studies involving human participants were in accordance with the World Medical Association’s Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.