Non-contrast 3.0T coronary magnetic resonance angiography: Image quality comparison of coronal balanced-turbo-field-echo, axial turbo-field-echo, and modified-Dixon sequences

Abstract

BACKGROUND

Non-contrast whole-heart coronary magnetic resonance angiography (CMRA) remains underutilized in clinical practice due to limited visualization of distal vessels and prolonged acquisition times.

AIM

To evaluate the performance of a coronal balanced turbo field echo (BTFE) sequence for CMRA at 3.0T in comparison with conventional axial turbo field echo (TFE) and modified Dixon (mDixon) sequences.

METHODS

Healthy young volunteers were prospectively enrolled from January 2025 to April 2025. All participants underwent coronary artery imaging using BTFE, TFE, and mDixon sequences. Subjective image quality was assessed based on the society of cardiovascular computed tomography 18-segment model using a four-point scale (1 = non-assessable to 4 = excellent) across the three major coronary arteries and their side branches. The assessability rate was defined as the percentage of segments receiving a score ≥ 2. Objective evaluation of the main coronary arteries included measurements of signal-to-noise ratio, contrast-to-noise ratio, vessel edge sharpness, and visible vessel length. Acquisition time was recorded for each sequence. Statistical comparisons among the three sequences were performed using the Friedman test and one-way repeated measures analysis of variance.

RESULTS

A total of 22 participants (mean age ± SD, 23 ± 2 years; 12 men) were included in the study. BTFE significantly outperformed both TFE and mDixon in subjective image quality scores across all coronary segments (all P < 0.05), except for the left main coronary artery, the proximal and mid segments of the left anterior descending artery (LAD), the proximal and mid segments of the left circumflex artery (LCX), the proximal segment of the right coronary artery, and ramus intermedius, for which no significant differences were observed between BTFE and mDixon (P > 0.05). The BTFE sequence achieved the highest overall assessability (99.3%, 282/284), surpassing mDixon (95.4%, 271/284) and TFE (87.0%, 247/284). BTFE yielded significantly higher signal-to-noise ratio, contrast-to-noise ratio, and vessel edge sharpness compared with both TFE and mDixon (all P < 0.001). The measured lengths of the LAD and right coronary artery were also significantly greater with BTFE (P < 0.05 for both). The LCX length was significantly longer on BTFE than on TFE (77.5 ± 10.5 mm vs 75.1 ± 8.3 mm; P = 0.017), but not significantly different from that on mDixon (77.5 ± 10.5 mm vs 75.7 ± 8.8 mm; P = 0.074). Additionally, BTFE demonstrated a shorter and more consistent acquisition time compared with both TFE and mDixon (6.13 ± 1.00 minutes vs 7.15 ± 2.02 minutes vs 7.17 ± 1.72 minutes; P < 0.05).

CONCLUSION

Coronal BTFE-based non-contrast whole-heart CMRA at 3.0T offers superior image quality and reduced acquisition time compared with conventional axial TFE and mDixon sequences.

Key Words: 3.0T; Magnetic resonance angiography; Coronary artery; Coronary artery disease; Image quality

Core Tip: This study presents a focused technical evaluation of a newly implemented coronal-plane balanced turbo field echo (BTFE) sequence for non-contrast-enhanced coronary magnetic resonance angiography at 3.0T. Through a prospective intra-individual comparison with conventional axial turbo field echo and modified Dixon sequences, the coronal BTFE protocol demonstrated superior image quality - particularly in distal coronary segments - and a shorter acquisition time. These findings indicate that coronal BTFE-based whole-heart coronary magnetic resonance angiography provides a more efficient and diagnostically robust alternative for non-contrast coronary imaging at 3.0T.

INTRODUCTION

Cardiovascular diseases, particularly coronary artery disease (CAD), remain the leading cause of mortality worldwide[1]. The current clinical assessment of CAD primarily relies on two imaging modalities: Invasive digital subtraction angiography[1] considered the diagnostic gold standard, and noninvasive coronary computed tomography angiography (CCTA). However, both techniques require iodinated contrast agents and expose patients to ionizing radiation, which poses potential risks. Non-contrast coronary magnetic resonance (MR) angiography (CMRA) has emerged as a safe, noninvasive, radiation-free alternative for imaging coronary artery stenosis without the need for iodinated contrast agents[2,3]. In addition to evaluating coronary arteries, CMRA can detect coronary artery aneurysms and assess coronary artery bypass grafts[4-6]. Furthermore, unlike CCTA, CMRA is not affected by calcification-induced artifacts[7,8].

Despite these advantages, the clinical adoption of CMRA remains limited due to technical challenges related to image quality and acquisition time. To address these limitations, various MR sequences have been developed and refined. The modified Dixon (mDixon) technique, based on water-fat separation, offers improved vessel-to-fat contrast and increased signal-to-noise ratio (SNR), along with enhanced tolerance to field inhomogeneity[9,10]. Previous studies have demonstrated that non-contrast CMRA using the mDixon technique at 3T provides excellent image quality and promising diagnostic performance[11-13]. Another widely used approach is the T2-prepared balanced steady-state free precession (SSFP) sequence, known for its intrinsically high blood signal and favorable blood-to-myocardium contrast[14]. Turbo field echo (TFE), a commonly implemented SSFP technique, is often employed in coronary imaging; however, it remains sensitive to flow artifacts, which can impair diagnostic confidence.

Recently, a coronal-plane implementation of a balanced TFE (BTFE) sequence has been proposed as a promising alternative to non-contrast CMRA[15]. This SSFP-based approach is designed to optimize acquisition efficiency and enhance image quality by improving SNR and reducing scan time. Although previous studies have compared BTFE with mDixon at 3T and demonstrated its diagnostic accuracy in patients with suspected CAD using CCTA as the reference standard, these investigations primarily focused on clinical populations and diagnostic outcomes. In contrast, direct head-to-head technical comparisons among BTFE, TFE, and mDixon sequences under conditions free from disease-related variability remain limited. Therefore, the objective of this prospective study was to address this gap by systematically comparing the performance of BTFE, TFE, and mDixon sequences for non-contrast whole-heart CMRA at 3.0T.

MATERIALS AND METHODS

Patients

This study was approved by the institutional review board of Beijing Anzhen Hospital, Capital Medical University, approval No. 2025232x. Written informed consent was obtained from all participants. Between January 2025 and April 2025, 22 healthy young volunteers were prospectively enrolled, all of whom underwent non-contrast, free-breathing, whole-heart CMRA. Each participant underwent scanning with three sequences, BTFE, TFE, and mDixon, on a 3.0T Philips system.

Image acquisition

CMRA examinations were performed using a 3.0T MR scanner (Gyroscan Intera, Philips Medical Systems, Best, Netherlands) equipped with a dedicated cardiac software package, a synergy cardiac coil, and a vectorcardiogram. Acquisition parameters for the BTFE, TFE, and mDixon sequences are summarized in Table 1.

Table 1 Non-contrast coronary magnetic resonance angiography acquisition parameters.

Parameter	BTFE	TFE	mDixon
No. slices	250	150	150
Field of view (mm × mm)	280 × 340	265 × 301	265 × 301
Matrix	208 × 234	212 × 201	176 × 201
Repetition time (millisecond)	Shortest	Shortest	Shortest
Echo time (millisecond)	Shortest	Shortest	Shortest
Flip angle (degrees)	70	20	10
Number of signal averages	1	1	1
Fat/blood suppression	Spectral-attenuated inversion recovery	Spectral-attenuated inversion recovery	Water-fat separation
Acquired voxel (mm3)	1.35 × 1.45 × 1.55	1.25 × 1.50 × 1.50	1.50 × 1.50 × 1.50
Reconstructed voxel (mm3)	0.75 × 0.75 × 0.75	0.75 × 0.75 × 0.75	0.75 × 0.75 × 0.75

BTFE: Balanced turbo field echo; TFE: Turbo field echo; mDixon: Modified Dixon.

Participants were positioned supine and head-first in the scanner. Survey images were obtained using the BTFE sequence to localize the heart. High-temporal-resolution cine imaging (100 phases per cardiac cycle) in the four-chamber view was then conducted. Subsequently, a BTFE sequence was used to determine the subject-specific trigger delay and the optimal data acquisition window, during which right coronary artery motion was minimal.

The same cardiac phase (diastole or systole) was selected for each participant across all three imaging protocols. A beam-shaped navigator was positioned at the dome of the right hemidiaphragm, with approximately one-third of the beam over the lung and two-thirds over the liver. The BTFE sequence was acquired in the coronal plane, whereas the TFE and mDixon sequences were acquired in the axial plane. All CMRA acquisitions were performed under free-breathing conditions. The total acquisition time varied according to navigator efficiency. Assuming 100% scan efficiency, the estimated acquisition times were 2 minutes 24 seconds for BTFE, 3 minutes 23 seconds for TFE, and 3 minutes 48 seconds for mDixon. For each CMRA scan, the participant’s heart rate and the actual acquisition time were recorded.

Imaging analysis

All examinations were independently evaluated by two radiologists with over 10 years of experience in cardiovascular imaging (Nan Zhang 15 years; Jia-Yi Liu 25 years). The names, dates, and sequences of each case were blinded to the readers. Image post-processing and evaluation were conducted using a Vitrea workstation (v8, Vital Images, Minnetonka, MN, United States). Image quality for all sequences was assessed from axial-plane images.

A total of 18 coronary artery segments were evaluated in each participant, following the society of cardiovascular computed tomography 18-segment model: Left main coronary artery (LM); left anterior descending artery LAD-proximal (pro), LAD-mid, LAD-distal (dis); first and second diagonal branches (D1, D2); left circumflex artery (LCX-pro, LCX-dis); first and second obtuse marginal branches (OM1, OM2); ramus intermedius; right coronary artery (RCA-pro, RCA-mid, RCA-dis); left and right posterior descending arteries (PDA); and left and right posterolateral branches (PLB). Because certain small vessels were present in only a subset of participants, D1 and D2 were pooled into a single diagonal branch (D) category, OM1 and OM2 into an OM category, left-PDA and right-PDA into a PDA category, and left-PLB and right-PLB into a PLB category to ensure adequate sample size for statistical analysis.

CMRA image quality was graded on a four-point scale: 1 = non-assessable (severe artifacts, poor vessel contrast); 2 = assessable (moderate artifacts, fair vessel contrast); 3 = assessable (minor artifacts, good vessel contrast); and 4 = assessable (no apparent artifacts, excellent vessel contrast)[16,17]. If either reader assigned a score of 1 (non-assessable), the segment was classified as non-assessable by consensus.

SNR and contrast-to-noise ratio (CNR) were measured on axial-plane images at four coronary locations: The LM, LAD-pro, LCX-pro, and RCA-pro. At each predefined measurement site, including the LM and the proximal segments (10 mm from the origin) of the RCA, LAD, and LCX, a circular region of interest (ROI) with an area of approximately 5 mm² was manually placed at the center of the vessel lumen to measure mean blood signal intensity. A second ROI of similar size was placed in the anterior chest wall musculature at the same axial level to assess background noise. ROI placement was standardized across sequences and participants, thereby avoiding partial-volume effects, proximity to the vessel wall, and regions affected by motion or flow artifacts. SNR was defined as SNR = SI_{coronary artery}/SD_musculature, where SI denotes signal intensity, and SD represents the standard deviation. CNR was defined as CNR = (SI_{coronary artery} - SI_musculature)/SD_musculature. Vessel edge sharpness (VES) was derived using a gradient-based algorithm in MATLAB (R2019a; MathWorks).

Coronary MRA images were further processed using curved planar reformation on the Vitrea post-processing workstation, and the visible vessel lengths of the LAD, LCX, and RCA were measured. The LAD length measurement included the LM segment.

The reproducibility of the image analysis was assessed through intra- and interobserver studies. To assess intraobserver variability, one reader (Nan Zhang) reviewed the images twice, with a two-week interval to reduce bias. For interobserver agreement, the same CMRA scans were independently reviewed by another reader (Jia-Yi Liu) and compared with the initial reader’s first set of measurements.

Statistical analysis

Data were tested for normality using the Shapiro-Wilk test. Normally distributed variables are reported as mean ± SD, while non-normally distributed data are presented as median (interquartile range). To compare measurements across the BTFE, TFE, and mDixon sequences, the Friedman test and one-way repeated measures analysis of variance were applied. For normally distributed variables, a one-way repeated-measures analysis of variance with post hoc Bonferroni correction was used. For non-normally distributed variables, the Friedman test was followed by pairwise Wilcoxon signed-rank tests with Bonferroni adjustment.

Interreader reproducibility for SNR, CNR, VES, and vessel length measurements was assessed using intraclass correlation coefficients (range: 0-1.00, with 0.81-1.00 indicating excellent agreement). Interreader agreement for image quality scoring was evaluated using κ statistics (range: 0-1.00, with κ ≥ 0.80 indicating excellent agreement). All statistical analyses were performed using SPSS version 27.0 (IBM, New York, NY, United States). A P value < 0.05 was considered statistically significant.

RESULTS

Participant characteristics

A total of 22 healthy participants (12 men; mean age 23 ± 2 years) were prospectively enrolled. The mean heart rate (HR) was 67 ± 7 beats/minutes, and the mean body mass index was 21.2 ± 1.4 kg/m². All participants completed non-contrast CMRA using the three sequences.

Assessment of image quality

Subjective image quality score and vessel assessability rate: As summarized in Table 2, visualization scores for distal coronary segments, including LAD-dis, LCX-dis, RCA-mid, RCA-dis, D, OM, PDA, and PLB, were significantly higher with BTFE than with TFE or mDixon (P < 0.05 for all). For proximal segments - including the LM, LAD-pro, LAD-mid, LCX-pro, RCA-pro, and ramus intermedius, BTFE yielded significantly higher scores than TFE (P < 0.05 for all), but showed no significant differences from mDixon (P > 0.05 for all) (Figures 1 and 2). The BTFE sequence achieved the highest overall vessel assessability rate (99.3%, 282/284), outperforming mDixon (95.4%, 271/284) and TFE (87.0%, 247/284). BTFE achieved 100% assessability in all major segments and side branches (D and OM), and demonstrated superior rates in distal segments, including PDA (94.4% vs 38.9% for TFE vs 83.3% for mDixon) and PLB (92.9% vs 57.1% vs 85.7%) (Tables 2 and 3).

Figure 1 Stacked bar charts show image quality scores of three coronary magnetic resonance angiography sequences. The stacked bar charts depict the distribution of subjective image quality scores (1-4) across 18 coronary segments for the three coronary magnetic resonance angiography sequences (balanced turbo field echo, turbo field echo, and modified Dixon). A: The stacked bar chart depict the distribution of subjective image quality scores (1-4) across left main coronary artery and the proximal, mid, and distal segments of the left anterior descending artery for the three coronary sequences; B: The stacked bar chart depict the distribution of subjective image quality scores (1-4) across the proximal and distal segments of the left circumflex artery for the three coronary sequences; C: The stacked bar chart depict the distribution of subjective image quality scores (1-4) across the proximal, mid, and distal segments of the right coronary artery for the three coronary sequences; D: The stacked bar chart depict the distribution of subjective image quality scores (1-4) across ramus intermediu, diagonal branch, obtuse marginal branch, posterior descending artery and posterolateral branch for the three coronary sequences. BTFE: Balanced turbo field echo; TFE: Turbo field echo; mDixon: Modified Dixon; LM: Left main artery; LAD-pro: Proximal left anterior descending artery; LAD-mid: Mid left anterior descending artery; LAD-dis: Distal left anterior descending artery; LCX-pro: Proximal left circumflex artery; LCX-dis: Distal left circumflex artery; RCA-pro: Proximal right coronary artery; RCA-mid: Mid right coronary artery; RCA-dis: Distal right coronary artery; RI: Ramus intermediu; D: The first and second diagonal branches (D1, D2); OM: The first and second obtuse marginal branches (OM1, OM2); PDA: Posterior descending arteries; PLB: Posterolateral branches.

Figure 2 Non-contrast coronary magnetic resonance angiography images show the three major coronary arteries across the balanced turbo field echo, turbo field echo, and modified Dixon sequences. The balanced turbo field echo (BTFE) sequence demonstrated superior image quality for all three arteries. In the right coronary artery, subjective image quality scores were 4 for BTFE, 2 for turbo field echo (TFE), and 3 for modified Dixon. The measured length of the left anterior descending artery was greater on BTFE images than on those acquired with TFE and modified Dixon. The lower contrast-to-noise ratio of the left circumflex artery on TFE likely explains its reduced measured length on that sequence. BTFE: Balanced turbo field echo; TFE: Turbo field echo; mDixon: Modified Dixon; LAD: Left anterior descending artery; LCX: Left circumflex artery; RCA: Right coronary artery.

Table 2 Participant characteristics and comparison of subjective image quality score at balanced turbo field echo, turbo field echo, and modified Dixon sequences.

Characteristic	BTFE	TFE	mDixon	Intergroups P value1	P value (adjusted P value)2	P value (adjusted P value)3	P value (adjusted P value)4
Sex	-	-	-	NA	NA	NA	NA
Male	12	12	12	-	-	-	-
Female	10	10	10	-	-	-	-
Age (year), mean ± SD	23 ± 2	23 ± 2	23 ± 2	NA	NA	NA	NA
HR (beats/minute), mean ± SD	67 ± 7	67 ± 7	67 ± 7	NA	NA	NA	NA
BMI, mean ± SD	21.2 ± 1.4	21.2 ± 1.4	21.2 ± 1.4	NA	NA	NA	NA
Image quality score
LM	4.0 (4.0-4.0)	3.0 (3.0-3.0)	4.0 (4.0-4.0)	< 0.001	< 0.001	0.177	0.009
LAD-pro	4.0 (4.0-4.0)	3.0 (3.0-4.0)	4.0 (4.0-4.0)	< 0.001	< 0.001	0.306	0.003
LCX-pro	4.0 (4.0-4.0)	3.0 (3.0-3.25)	4.0 (4.0-4.0)	< 0.001	< 0.001	0.583	< 0.001
RCA-pro	4.0 (4.0-4.0)	3.0 (3.0-3.25)	4.0 (4.0-4.0)	< 0.001	< 0.001	0.084	< 0.001
LAD-mid	4.0 (4.0-4.0)	3.0 (2.75-3.0)	4.0 (3.875-4.0)	< 0.001	< 0.001	0.489	< 0.001
RCA-mid	4.0 (4.0-4.0)	3.0 (2.5-3.0)	3.75 (2.875-4.0)	< 0.001	< 0.001	0.024	< 0.001
LAD-dis	4.0 (3.875-4.0)	3.0 (2.0-3.0)	3.0 (3.0-4.0)	< 0.001	< 0.001	< 0.001	0.009
LCX-dis	4.0 (3.0-4.0)	3.0 (2.0-3.0)	3.0 (2.875-3.25)	< 0.001	< 0.001	0.03	< 0.001
RCA-dis	4.0 (3.375-4.0)	3.0 (2.0-3.0)	3.0 (2.75-4.0)	< 0.001	< 0.001	0.03	< 0.001
RI	3.5 (2.88-4.0)	2.0 (2.0-2.63)	3.0 (2.0-4.0)	0.009	0.033	0.156	0.21
D	3.0 (3.0-4.0)	2.0 (1.0-3.0)	3.0 (2.0-3.375)	< 0.001	< 0.001	0.009	0.006
OM	4.0 (3.0-4.0)	2.0 (2.0-3.0)	3.0 (2.0-4.0)	< 0.001	< 0.001	0.015	0.09
PDA	3.0 (2.0-4.0)	1.0 (1.0-2.0)	2.0 (2.0-3.0)	< 0.001	< 0.001	0.033	0.006
PLB	3.0 (3.0-4.0)	2.0 (1.0-2.0)	3.0 (2.0-3.0)	< 0.001	0.006	0.042	0.018

¹The Friedman test was used to compare characteristics among balanced turbo field echo, turbo field echo, and modified Dixon.

²Balanced turbo field echo vs turbo field echo.

³Balanced turbo field echo vs modified Dixon.

⁴Turbo field echo vs modified Dixon.

NA: Not available; BTFE: Balanced turbo field echo; TFE: Turbo field echo; mDixon: Modified Dixon; LM: Left main artery; LAD-pro: Proximal left anterior descending artery; LAD-mid: Mid left anterior descending artery; LAD-dis: Distal left anterior descending artery; LCX-pro: Proximal left circumflex artery; LCX-dis: Distal left circumflex artery; RCA-pro: Proximal right coronary artery; RCA-mid: Mid right coronary artery; RCA-dis: Distal right coronary artery; RI: Ramus intermediu; D: The first and second diagonal branches (D1, D2); OM: The first and second obtuse marginal branches (OM1, OM2); PDA: Posterior descending arteries; PLB: Posterolateral branches.

Table 3 Comparison of vessel assessability rates of non-contrast coronary magnetic resonance angiography sequences at balanced turbo field echo, turbo field echo, and modified Dixon in 22 participants, n (%).

Segment	No. of all segments	No. of assessable segment (score ≥ 2)
		BTFE	TFE	mDixon
Overall	284	282 (99.3)	247 (87.0)	271 (95.4)
LM	22	22 (100)	22 (100)	22 (100)
LAD-pro	22	22 (100)	22 (100)	22 (100)
LCX-pro	22	22 (100)	22 (100)	22 (100)
RCA-pro	22	22 (100)	22 (100)	22 (100)
LAD-mid	22	22 (100)	22 (100)	22 (100)
RCA-mid	22	22 (100)	21 (95.5)	21 (95.5)
LAD-dis	22	22 (100)	22 (100)	22 (100)
LCX-dis	22	22 (100)	20 (90.9)	22 (100)
RCA-dis	22	22 (100)	20 (90.9)	21 (95.5)
RI	10	10 (100)	9 (90.0)	9 (90.0)
D	24	24 (100)	14 (58.3)	20 (83.3)
OM	20	20 (100)	16 (80)	19 (95.0)
PDA	18	17 (94.4)	7 (38.9)	15 (83.3)
PLB	14	13 (92.9)	8 (57.1)	12 (85.7)

BTFE: Balanced turbo field echo; TFE: Turbo field echo; mDixon: Modified Dixon; LM: Left main artery; LAD-pro: Proximal left anterior descending artery; LAD-mid: Mid left anterior descending artery; LAD-dis: Distal left anterior descending artery; LCX-pro: Proximal left circumflex artery; LCX-dis: Distal left circumflex artery; RCA-pro: Proximal right coronary artery; RCA-mid: Mid right coronary artery; RCA-dis: Distal right coronary artery; RI: Ramus intermediu; D: The first and second diagonal branches (D1, D2); OM: The first and second obtuse marginal branches (OM1, OM2); PDA: Posterior descending arteries; PLB: Posterolateral branches.

Quantitative image quality assessment and acquisition time: BTFE significantly outperformed both TFE and mDixon across all quantitative metrics (P < 0.001 for all) (Table 4). Mean SNR values for BTFE ranged from 137 to 145, compared with 73 to 89 for TFE and 94 to 109 for mDixon (P < 0.001). Mean CNR values ranged from 75 to 84 for BTFE, 15 to 31 for TFE, and 35 to 50 for mDixon (P < 0.001). Mean VES was also significantly higher with BTFE, ranging from 257 to 284, compared with 141 to 150 for TFE and 77 to 82 for mDixon (P < 0.001).

Table 4 Quantitative image quality metrics and mean acquisition time for balanced turbo field echo, turbo field echo, and modified Dixon sequences, mean ± SD.

Parameter	BTFE	TFE	mDixon	Intergroups P value1	P value (adjusted P value)2	P value (adjusted P value)3	P value (adjusted P value)4
SNR
LM	145.0 ± 12.9	88.68 ± 8.2	109.4 ± 12.2	< 0.001	< 0.001	< 0.001	< 0.001
LAD-pro	139.7 ± 13.6	83.2 ± 7.9	109.4 ± 12.6	< 0.001	< 0.001	< 0.001	< 0.001
LCX-pro	137.2 ± 14.5	80.4 ± 6.3	104.8 ± 12.5	< 0.001	< 0.001	< 0.001	< 0.001
RCA-pro	135.4 ± 8.4	73.3 ± 9.2	94.3 ± 12.0	< 0.001	< 0.001	< 0.001	< 0.001
CNR	-	-	-	< 0.001	< 0.001	< 0.001	< 0.001
LM	83.7 ± 13.6	30.7 ± 6.4	50.2 ± 10.6	< 0.001	< 0.001	< 0.001	< 0.001
LAD-pro	78.2 ± 12.4	25.1 ± 5.8	50.1 ± 11.7	< 0.001	< 0.001	< 0.001	< 0.001
LCX-pro	75.7 ± 14.0	22.2 ± 7.1	45.5 ± 14.0	< 0.001	< 0.001	< 0.001	< 0.001
RCA-pro	74.1 ± 12.1	15.4 ± 6.7	35.1 ± 8.6	< 0.001	< 0.001	< 0.001	< 0.001
VES5	-	-	-	< 0.001	< 0.001	< 0.001	< 0.001
LM	257.3 ± 31.0	150.2 ± 19.3	80.5 ± 14.0	< 0.001	< 0.001	< 0.001	< 0.001
LAD-pro	266.1 ± 34.9	149.6 ± 14.2	82.1 ± 10.8	< 0.001	< 0.001	< 0.001	< 0.001
LCX-pro	276.6 ± 33.9	141.1 ± 13.0	76.9 ± 7.4	< 0.001	< 0.001	< 0.001	< 0.001
RCA-pro	284.1 ± 35.8	143.4 ± 13.5	79.7 ± 9.8	< 0.001	< 0.001	< 0.001	< 0.001
Vessel length (mm)
LAD	126 ± 10	114 ± 17	118 ± 17	< 0.001	< 0.001	0.016	0.017
LCX	77 ± 10	75 ± 8	76 ± 9	0.002	0.017	0.074	0.459
RCA	122 ± 7	120 ± 7	121 ± 7	< 0.001	0.002	0.043	0.038
Acquisition time (minute)	6.13 ± 1.00	7.15 ± 2.02	7.17 ± 1.72	0.015	0.025	0.035	1.000

¹The one-way repeated measures analysis of variance was used to compare characteristics among balanced turbo field echo, turbo field echo, and modified Dixon.

²Balanced turbo field echo vs turbo field echo.

³Balanced turbo field echo vs modified Dixon.

⁴Turbo field echo vs modified Dixon.

⁵Vessel edge sharpness was computed as the maximum normalized gradient of the signal-intensity profile perpendicular to the vessel wall, using a gradient-based algorithm implemented in Matrix Laboratory (R2019a; MathWorks).

SNR: Signal-to-noise ratio; CNR: Contrast-to-noise ratio; VES: Vessel edge sharpness; BTFE: Balanced turbo field echo; TFE: Turbo field echo; mDixon: Modified Dixon; LM: Left main artery; LAD: Left anterior descending artery; LCX: Left circumflex artery; RCA: Right coronary artery; LAD-pro: Proximal left anterior descending artery; LCX-pro: Proximal left circumflex artery; RCA-pro: Proximal right coronary artery.

The measured lengths of the LAD and RCA were significantly greater with BTFE compared to the other sequences (P < 0.005 for all). The LCX length was significantly longer on BTFE than on TFE (77.5 ± 10.5 mm vs 75.1 ± 8.3 mm; P = 0.017), but not significantly different from mDixon (77.5 ± 10.5 mm vs 75.7 ± 8.8 mm; P = 0.074).

BTFE CMRA also demonstrated a shorter mean acquisition time and reduced variability in imaging time compared with TFE and mDixon (6.13 ± 1.00 minutes vs 7.15 ± 2.02 minutes vs 7.17 ± 1.72 minutes; P < 0.005).

Agreements of scoring and measurements: Inter- and intra-observer reproducibility was excellent for all assessments, including image quality scoring, quantitative image quality metrics (SNR, CNR, VES), and vessel length measurements across the BTFE, TFE, and mDixon sequences. κ values for categorical assessments and intraclass correlation coefficients for continuous variables are summarized in Tables 5 and 6.

Table 5 Inter- and intra-observer agreements in the scoring of image quality among three different sequences.

Score of vessel segments	Reader 1	Reader 2	Interobserver kappa (95%CI)	Reader 1 repeat	Intraobserver kappa (95%CI)
LM	4.0 (3.0-4.0)	4.0 (3.0-4.0)	0.887 (0.763-1.012)	4.0 (3.0-4.0)	0.924 (0.819-1.028)
LAD-pro	4.0 (3.75-4.0)	4.0 (3.75-4.0)	0.835 (0.679-0.991)	4.0 (4.0-4.0)	0.874 (0.734-1.013)
LCX-pro	4.0 (3.0-4.0)	4.0 (3.0-4.0)	0.901 (0.792-1.010)	4.0 (3.0-4.0)	0.933 (0.843-1.024)
RCA-pro	4.0 (3.0-4.0)	4.0 (3.0-4.0)	0.886 (0.776-0.996)	4.0 (3.0-4.0)	0.914 (0.817-1.010)
LAD-mid	4.0 (3.0-4.0)	4.0 (3.0-4.0)	0.875 (0.766-0.985)	4.0 (3.0-4.0)	0.900 (0.802-0.998)
RCA-mid	4.0 (3.0-4.0)	3.0 (3.0-4.0)	0.887 (0.797-0.978)	3.0 (3.0-4.0)	0.906 (0.824-0.988)
LAD-dis	3.0 (3.0-4.0)	3.0 (3.0-4.0)	0.853 (0.743-0.963)	3.0 (3.0-4.0)	0.872 (0.768-0.976)
LCX-dis	3.0 (2.75-4.0)	3.0 (3.0-4.0)	0.859 (0.753-0.965)	3.0 (3.0-4.0)	0.877 (0.777-0.977)
RCA-dis	3.0 (3.0-4.0)	3.0 (2.0-4.0)	0.872 (0.774-0.969)	3.0 (2.0-4.0)	0.888 (0.796-0.980)
RI	3.0 (2.0-4.0)	3.0 (2.0-3.25)	0.865 (0.740-0.991)	3.0 (2.0-4.0)	0.904 (0.797-1.011)
D	3.0 (2.0-3.0)	3.0 (2.0-3.0)	0.924 (0.865-0.984)	3.0 (2.0-3.0)	0.950 (0.901-0.998)
OM	3.0 (2.0-4.0)	3.0 (2.0-4.0)	0.891 (0.812-0.970)	3.0 (2.0-4.0)	0.938 (0.879-0.998)
PDA	2.0 (1.0-3.0)	2.0 (1.0-3.0)	0.910 (0.836-0.984)	2.0 (1.0-3.0)	0.964 (0.915-1.014)
PLB	3.0 (2.0-3.0)	3.0 (2.0-3.0)	0.884 (0.787-0.981)	3.0 (2.0-3.0)	0.954 (0.892-1.016)

LM: Left main artery; LAD-pro: Proximal left anterior descending artery; LAD-mid: Mid left anterior descending artery; LAD-dis: Distal left anterior descending artery; LCX-pro: Proximal left circumflex artery; LCX-dis: Distal left circumflex artery; RCA-pro: Proximal right coronary artery; RCA-mid: Mid right coronary artery; RCA-dis: Distal right coronary artery; RI: Ramus intermediu; D: The first and second diagonal branches (D1, D2); OM: The first and second obtuse marginal branches (OM1, OM2); PDA: Posterior descending arteries; PLB: Posterolateral branches; CI: Confidence interval.

Table 6 Inter- and intra-observer agreements in the quantitative image quality and vessel length among three different sequences, mean ± SD.

Characteristic	Reader 1	Reader 2	Interobserver ICC (95%CI)	Reader 1 repeat	Intraobserver ICC (95%CI)
SNR
LM	114.4 ± 25.9	115.6 ± 23.0	0.957 (0.931-0.974)	114.4 ± 24.0	0.989 (0.982-0.993)
LAD-pro	110.8 ± 25.9	112.9 ± 30.7	0.966 (0.943-0.979)	111.4 ± 23.7	0.985 (0.976-0.991)
LCX-pro	107.4 ± 26.1	107.3 ± 22.4	0.956 (0.930-0.973)	108.4 ± 23.6	0.986 (0.977-0.991)
RCA-pro	103.6 ± 27.6	101.4 ± 26.6	0.975 (0.959-0.985)	103.8 ± 24.6	0.989 (0.968-0.995)
CNR
LM	54.9 ± 24.2	57.2 ± 28.7	0.951 (0.918-0.970)	55.9 ± 28.0	0.973 (0.956-0.983)
LAD-pro	51.2 ± 24.1	51.1 ± 20.2	0.948 (0.916-0.968)	52.1 ± 21.8	0.984 (0.973-0.990)
LCX-pro	47.8 ± 25.1	50.1 ± 28.8	0.952 (0.920-0.971)	48.8 ± 28.1	0.973 (0.957-0.984)
RCA-pro	41.5 ± 26.3	43.3 ± 27.0	0.971 (0.951-0.982)	43.3 ± 27.7	0.982 (0.967-0.990)
VES
LM	162.7 ± 76.6	155.7 ± 65.9	0.951 (0.917-0.971)	159.6 ± 75.4	0.983 (0.972-0.989)
LAD-pro	166.0 ± 79.8	159.0 ± 69.2	0.958 (0.927-0.975)	162.9 ± 78.2	0.984 (0.974-0.990)
LCX-pro	164.9 ± 86.5	161.5 ± 77.2	0.979 (0.966-0.987)	161.2 ± 81.0	0.985 (0.975-0.991)
RCA-pro	169.1 ± 88.9	162.0 ± 82.7	0.985 (0.964- 0.992)	160.9 ± 84.6	0.989 (0.949-0.996)
Vessel length (mm)
LAD	119.2 ± 16.0	119.3 ± 16.0	0.997 (0.994-0.998)	118.7 ± 16.2	0.999 (0.999-1.000)
LCX	76.1 ± 9.1	75.9 ± 9.8	0.984 (0.975-0.990)	76.2 ± 9.7	0.989 (0.982-0.993)
RCA	121.4 ± 7.2	122.4 ± 7.2	0.978 (0.908-0.991)	121.5 ± 7.2	0.996 (0.994-0.998)

ICC: Intraclass correlation coefficients; CI: Confidence interval; SNR: Signal-to-noise ratio; CNR: Contrast-to-noise ratio; VES: Vessel edge sharpness; LM: Left main artery; LAD: Left anterior descending artery; LCX: Left circumflex artery; RCA: Right coronary artery; LAD-pro: Proximal left anterior descending artery; LCX-pro: Proximal left circumflex artery; RCA-pro: Proximal right coronary artery.

DISCUSSION

This study compared CMRA image quality and acquisition time across BTFE, TFE, and mDixon sequences. The main findings are as follows: (1) Radiologist evaluations revealed significantly improved coronary visualization with BTFE, particularly in distal segments, compared with the other sequences; (2) The coronal-plane BTFE CMRA sequence achieved higher SNR, CNR, and VES, along with a shorter acquisition time, relative to both TFE and mDixon; and (3) BTFE provided superior delineation of the LAD and RCA compared with the other two techniques.

Although the mDixon technique has gained broader application in 3.0T CMRA due to its robust fat suppression and improved tolerance to B0 inhomogeneity[9], it typically requires longer acquisition times and often relies on multiphase data acquisition within the same respiratory window to ensure accurate water-fat separation[15]. These requirements may reduce its efficiency for routine coronary imaging. TFE, while widely used in clinical settings, remains sensitive to flow-related artifacts and may exhibit less consistent blood-myocardium contrast, particularly in distal and branch vessels.

In contrast, the coronal BTFE sequence offers higher intrinsic SNR efficiency and more stable magnetization during free-breathing acquisition. This results in significantly improved SNR, CNR, and VES, along with shorter and more consistent scan times in our study cohort. These characteristics collectively account for the superior overall performance of BTFE compared with both TFE and mDixon in this direct technical comparison.

In this cohort of healthy volunteers, the BTFE sequence demonstrated superior image quality, particularly in visualizing distal coronary artery segments and side branches - regions that remain challenging in noninvasive coronary evaluation. Notably, BTFE achieved an assessability rate of 99.3%, with 100% visualization of all major coronary trunks and side branches (including Ds and OMs) and maintained high visualization rates in the distal PDA and PLB. These findings are clinically meaningful and suggest that BTFE may address key limitations of current non-contrast CMRA techniques. The sequence demonstrates robustness and consistency, supporting its potential utility as a preferred option in comprehensive CMRA protocols.

Quantitative analysis demonstrated that the coronal plane BTFE sequence significantly increased the visualized lengths of the LAD and RCA compared with the mDixon sequences, whereas measurements of the LCX showed no significant differences. The discrepancy observed in LAD measurements may be attributed to its distinct anatomical characteristics: Approximately 67%-98% of myocardial bridges are located in the LAD, most commonly in its proximal and mid segments[18]. When the CNR falls below a critical threshold, the signal intensity of the vessel lumen becomes indistinguishable from the surrounding myocardium, complicating accurate delineation. The higher CNR achieved with the BTFE sequence addresses this limitation by enhancing the contrast between the LAD lumen and adjacent myocardial tissue, thereby increasing its measurable length. Furthermore, because the RCA generally extends more distally than the LCX, the superior distal segment visualization provided by BTFE likely contributed to the longer apparent RCA length, consistent with radiologists’ subjective image quality assessments.

This study differs in scope and study population from a recent work by Yuan et al[15], who evaluated the coronal-plane BTFE sequence in patients with suspected CAD, using CCTA as the reference standard. Their investigation focused on diagnostic performance, including sensitivity and specificity for stenosis detection. It demonstrated superior image quality and diagnostic accuracy of BTFE compared with mDixon in a clinical setting. In contrast, the present study was conducted in healthy volunteers and focused on a technical comparison of three noncontrast CMRA sequences - BTFE, TFE, and mDixon - at 3.0T. Unlike Yuan et al[15] included TFE as a conventional sequence not evaluated in their work and conducted a more comprehensive technical assessment encompassing quantitative metrics (SNR, CNR, and VES), subjective image quality scores for both major coronary trunks and side branches, vessel assessability, and total scan time. These findings not only complement those of Yuan et al[15] by confirming the superior performance of BTFE over TFE and mDixon but also extend their conclusions by demonstrating excellent image quality and high assessability of distal segments and side branches. This is particularly relevant given that limited visualization of distal segments and side branches has long been a major constraint in the broader clinical adoption of CMRA.

The robust performance of BTFE in anatomically and diagnostically challenging segments suggests its potential to overcome current technical limitations and enhance diagnostic confidence in comprehensive coronary assessment. Accordingly, further investigation of patients with suspected CAD is warranted, particularly to evaluate the diagnostic utility of side branch imaging. Collectively, these findings highlight the clinical potential of BTFE as a technically reliable and diagnostically versatile sequence for non-contrast coronary MRA.

Additionally, ECG trigger delay and data acquisition windows were individually optimized for each subject based on the RCA’s relative quiescent phase during coronary MRA acquisition. Despite this tailored approach, RCA images exhibited lower SNR and CNR compared with those of other coronary arteries. This discrepancy is likely due to residual vessel motion and the RCA’s distinct flow dynamics, which may compromise image stability even during nominally motion-free periods.

Additionally, these observations indicate that in CMRA, performing high-temporal-resolution cine imaging under conditions of low HR variability to determine the trigger delay and data acquisition window is critical for ensuring the image quality of subsequent CMRA. For patients with a baseline HR exceeding 75 bpm, which may lead to suboptimal image quality, sublingual administration of nitroglycerin can be used to reduce the HR to the target range of 55-65 bpm[19,20]. Lu et al[12] demonstrated that sublingual administration of nitroglycerin significantly enhances coronary artery conspicuity and overall image quality, particularly in distal segments, and improves the diagnostic performance of non-contrast coronary MRA for CAD. Additionally, it is noteworthy that autonomic nervous system-mediated changes in breathing patterns during sleep can lead to increased diaphragm displacement, which results in decreased navigator gating efficiency and a significant prolongation of adequate acquisition time. Therefore, patients should be prevented from falling asleep during the scan.

This study has several limitations. First, the sample size was relatively small due to its single-center design. Inclusion of a larger number of participants, particularly through multicenter collaboration, is desirable. Second, the inclusion criteria were limited to young, healthy individuals, excluding elderly participants and those at high risk for CAD. Consequently, the applicability of these findings to the broader clinical population requires further validation. Future research is planned as a prospective, multicenter cohort study with a larger sample size and broader demographic stratification to assess the generalizability of this technique.

CONCLUSION

Compared with uncontrast CMRA using TFE and mDixon sequences, the BTFE-based approach demonstrated significantly superior performance in subjective image quality, SNR, CNR, VES, and measurable vessel length. These findings suggest that BTFE-CMRA may serve as a promising alternative for the noninvasive clinical evaluation of CAD.