High resolution contrast-enhanced MRA (CE-MRA) is performed during the 1st pass of gadolinium contrast. Typically, the full contrast dose is injected at 1.5 – 2.0 mL/s, followed by saline flush at the same rate. Centric image acquisition begins near peak enhancement using fluoroscopic triggering or a test bolus. Any temporal variation in vascular enhancement during acquisition (e.g., “peaked” bolus with subsequent fall off) causes blurring and ringing artifacts in the images. The purpose of this study was to individually “tailor” the contrast bolus enhancement profile to be constant over the scan time, with the ultimate goal of improving CE-MRA image quality.

With local IRB approval, 20 subjects were randomized to receive either standard or “tailored” injection (Medrad Spectris Solaris, Bayer, Whippany, NJ) of single-dose (0.1 mmol/kg) gadoteridol (Bracco, Princeton, NJ), and resulting signal intensity (SI) profiles were measured for 100 sec to capture pre-Gd baseline, 1st pass, and recirculation. Prior to each full injection, a test bolus was administered. Standard injections consisted of non-diluted, single-dose contrast (less test bolus volume) injected at our institutional standard 1.6 mL/s. Tailored injections utilized a novel optimization algorithm to determine the best multi-phase (≤ 3 phases) injection of 38 mL diluted (single-dose diluted with saline to 40mL) contrast. The test boluses consisted of 1 mL pure contrast @ 1.6 mL/s (standard), or 2 mL diluted contrast @ 2 mL/s (tailored). The optimization utilized linear prediction of contrast concentration profiles (based on the test bolus response) for arbitrary injection schemes, and targeted a 20 s SI plateau duration. This was implemented in MATLAB (Mathworks, Natick, MA). All contrast injections were followed by 25-30 mL saline flush at the final contrast injection rate.

Signal intensity profiles for all injections were measured (3T Ingenia, Philips, Best, the Netherlands) using an oblique sagittal, multi-dynamic, thick slice 3D T1-FFE acquisition covering the entire aorta to eliminate inflow effects. Time-intensity curves were gathered from a single ROI placed in the supra-celiac abdominal aorta. After normalization to baseline, maximum SI and full width at 80% max (FWM₈₀) SI were measured. Test bolus MR parameters were: TR/TE/α = 2ms/0.8ms/7^o, temporal resolution ~ 1 s; and full bolus: TR/TE/α = 3.5ms/1.4ms/30^o, temporal resolution ~ 1.7 s. Spatial resolution was uniformly 1.8 x.1.8 x 7 mm³.

Figure 1 illustrates the bolus tailoring concept for a single subject, demonstrating timing bolus results, calculated 3-phase injection scheme, and resultant full bolus. Ranges for each phase (a-c) of the patient-specific triphasic tailored bolus volumes and rates were: a) 8.3 + 3 mL @ 3.1 + 0.6 mL/s; b) 16.4 + 7.7 mL @ 1.4 + 0.2 mL/s; and c) 10.0 + 5.0 mL @ 1.3 + 0.2 mL/s.

The single-phase injections were markedly peaked, demonstrating a relatively short maximum SI duration, with mean FWM₈₀ = 8.8 + 3.8 s (Figure 2, left). The bolus-tailoring prediction algorithm worked well to achieve the desired uniform SI plateau with significantly increased duration; mean FWM₈₀ = 23.7 + 2.5 s (p < 0.01) (Figure 2, right). Only a relatively small mean signal loss (19.5%, p = 0.05) was observed to achieve this much longer bolus duration (Figures 3 and 4).

Using a predictive algorithm with a small contrast test bolus allows for patient-specific enhancement profile “shaping” to achieve a longer and more uniform plateau than does a typical single-phase injection. This prolonged SI plateau comes at the expense of a modest SI decrease (i.e., nearly 3x plateau duration while maintaining >80% of peak SI). This is largely explained by T1 and T2* mediated sub-linearity of SI vs. contrast concentration such that significant decrease in contrast injection rates lead to only minimal SI decrease (1,2).

Of note, we found it necessary to dilute the contrast for bolus tailoring, as calculated injection rates for undiluted contrast, particularly the 2nd and 3rd phase, were quite low (often ~ 0.5-0.8 mL/s). Because the test bolus is performed at a fixed injection rate, system linearity can only be assumed for that rate. In preliminary tests without diluting, we found that slower bolus phases arrived later than predicted, distorting the expected plateau phase (Figure 5). This was largely remedied in the study by diluting the contrast as described, keeping injection rates higher where the system behaved more linearly.

Bolus tailoring for CE-MRA can be readily achieved and has significant potential to improve image quality. This suggests the conventional method of injecting Gd contrast at a relatively fast fixed rate is flawed. Future work will evaluate the impact of tailored boluses on CE-MRA image quality.