Adaptive vessel tracking: Automated computation of vessel trajectories for improved efficiency in 2D coronary MR angiography

Abstract

A new method was investigated for improving the efficiency of ECG-gated coronary magnetic resonance angiography (CMRA) by accurate, automated tracking of the vessel motion over the cardiac cycle. Vessel tracking was implemented on a spiral gradient-echo pulse sequence with sub-millimeter in-plane spatial resolution as well as high image signal to noise ratio. Breath hold 2D CMRA was performed in 18 healthy adult subjects (mean age 46 ؎ 14 years). Imaging efficiency, defined as the percentage of the slices where more than 30 mm of the vessel is visualized, was computed in multi-slice spiral scans with and without vessel tracking. There was a significant improvement in the efficiency of the vessel tracking sequence compared to the multi-slice sequence (56% vs. 32%, P < 0.001). The imaging efficiency increased further when the true motion of the coronary arteries (determined using a cross correlation algorithm) was used for vessel tracking as opposed to a linear model for motion (71% vs. 57%, P < 0.05). The motion of the coronary arteries was generally found to be linear during the systolic phase and nonlinear during the diastolic phase. The use of subject-tailored, auto- CORONARY MAGNETIC RESONANCE Angiography (CMRA) remains challenging due to constraints imposed by respiratory and cardiac motion. Acquisition of the images during an end-expiratory breath hold is commonly employed to minimize respiratory artifacts, while electrocardiography (ECG) gating can effectively freeze cardiac motion. Edelman and Manning (1,2) were the first to describe breath-held two-dimensional (2D) CMRA acquired during mid-diastole, using an ECGgated segmented k-space gradient echo sequence. The use of newer, faster acquisition schemes that minimize arterial blurring, has significantly improved image quality (3,4).The two most commonly employed schemes are the acquisition of the same image plane over the entire cardiac cycle (traditional "cine") and the acquisition, during mid-diastole, of multiple slices with contiguous cardiac phases. However, both these methods require that segments of the coronary artery be constrained within the planes of the prescribed slices and do not make any adjustments to synchronize image acquisition with the actual changing vessel location. Since there is substantial motion of the coronary arteries (on the order of 2 cm or more) during the cardiac cycle, visualization is typically limited to a small fraction (approximately 20% (5)) of the total images acquired in each breath-hold. This implies repeated breath-hold acquisitions to ensure adequate anatomic coverage; however this is often ill tolerated in patients with coronary artery disease (6).In a previous study, Foo et al (7) proposed a method for improving the imaging efficiency called vessel tracking. A preparatory "cine" acquisition on a plane orthogonal to the CMRA imaging plane, where the cross-section of the arteries are depicted, was used to estimate the maximum excursion of the coronary arteries. By assuming a linear vessel motion across the cardia...

Displacement and velocity of the coronary arteries: cardiac and respiratory motion

“…Individual patient variability affects the shape and duration of these features. Displacement of the proximal RCA in 12 subjects as a function of the cardiac phase [14]. These measurements were made from 2-D breath held magnetic resonance images, and correlate well with results obtained in this study as shown in Fig.…”

“…A plot of RCA displacement as a function of the cardiac cycle found in [14] (see Fig. 5) is very similar to Fig.…”

Coronary Vessel Wall 3-T MR Imaging with Time-resolved Acquisition of Phase-Sensitive Dual Inversion-Recovery (TRAPD) Technique: Initial Results in Patients with Risk Factors for Coronary Artery Disease

“…No statistical difference (P = .06) was found between the body mass index of subjects with CAD risk factors (mean, 25 6 5; range, [16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34] and healthy subjects (mean, 22 6 4; range, [20][21][22][23][24][25][26][27][28][29][30]. Subjects with CAD risk factors were significantly older than healthy subjects (P , .001), while there was no significant difference in age between men and women with CAD risk factors (P = .7).…”

“…Previously proposed solutions-including vessel tracking (15)(16)(17)(18)(19), subject-specific acquisition windows (20), and adaptive trigger delays (21-23)-cannot guarantee preservation of the orthogonality required for optimal two-dimensional wall imaging and accurate quantitative assessment. Another challenge facing successful vessel wall imaging is the inherent requirement of blood signal nulling.…”

Variable‐density adaptive imaging for high‐resolution coronary artery MRI

“…In contrast, values listed in Table 4 refer to actual coronary displacement. On the other hand, recent studies using slice-following techniques demonstrated significantly improved navigator-echo efficiencies (տ70%) (28). Further study is therefore required to elucidate the exact nature of the data selected by the two approaches.…”