Purpose

Dynamic contrast-enhanced (DCE)-MRI is used to quantify organ perfusion abnormalities in many diseases¹. The purpose of this study was to investigate the accuracy of a new method for measuring the arterial input function (AIF) in DCE-MRI. This method combines phase-versus-time measurements with pre- and post-DCE Modified Look-Locker Inversion Recovery (MOLLI)² T₁ measurements. This method, denoted “Phase+MOLLI”, was compared to an older method which uses only phase measurements (“Phase-only”)^3-4.

Methods

Two MOLLI prototype acquisitions were used for T₁ measurements: “MOLLI-Blood” – optimized for higher gadolinium concentrations ([Gd]) found in blood (>0.5 mM) with inversion point-spacing of 175ms; “MOLLI-Brain” – optimized for lower [Gd] found in tissue, with inversion point-spacing of 750ms. [Gd] was quantified using a calibration curve for the combination of these two methods (“MOLLI-Combination”) into a single fit, where MOLLI-Brain was used to measure low [Gd] and MOLLI-Blood was used for higher [Gd] measurements. Seven DCE-MRI experiments were performed on a thin-tubing aqueous multimodality DCE perfusion flow phantom (Shelley Medical) in a 3T MRI system (Magnetom Trio; Siemens) with 32 channel head coil and clinical DCE-MRI protocol (FLASH: TR=6.5ms, TE₁/TE₂=1.65/3.85, flip angle=30º). The input flow rate was set to 200 mL/min to mimic blood flow in the human brain. A small amount of Gd (Gadovist® 1.0, Bayer) was added to water initially so the baseline T₁ was similar to blood. MOLLI T₁ measurements were performed before and after DCE acquisition. The AIF was measured by the 1) “Phase-only” method: change in phase as a function of time was used to compute the AIF^3-4; and 2) “Phase+MOLLI” method: [Gd] during the post-injection steady-state washout (AIF_w) was calculated from MOLLI T₁ values and known T₁ relaxivity. This value of AIF_w was used to translate the Phase-only AIF in the vertical (concentration) direction until the washout part of the curve equalled AIF_w, and setting the baseline to zero. A “gold-standard” AIF_w was also obtained by sampling liquid from the phantom pre- and post-DCE, and measuring T₁ with inversion recovery (EPI-IR). MOLLI was also performed on this sample. A paired t-test was conducted to compare AIF_w values from the Phase-only and Phase+MOLLI methods. The correlation between the [Gd] values from MOLLI and EPI-IR was also determined. To investigate flow effects, the DCE phantom flow was stopped and MOLLI was performed; [Gd] from these two flow states were compared.

Results

To determine if MOLLI was sensitive to flow for different [Gd] values, 1/T₁ was computed in both static and flowing water. Fig. 1 shows significant differences (all P<0.001) were observed before and after DCE for both MOLLI-Brain and MOLLI-Blood. To minimize flow effects in DCE acquisitions, MOLLI-Combination was used as it's the least sensitive to flow. Fig. 2 shows these effects on [Gd] from MOLLI-Blood (P<0.0001), MOLLI-Brain (P=0.007), and MOLLI-Combination (P=0.42) in flowing and static water. To verify MOLLI-Combination was accurate in determining [Gd], it was compared with gold-standard EPI-IR where good correlation was observed (P<0.0001, r²=0.9899) (Fig. 3). In the DCE acquisition, a significant difference was observed between [Gd] from the Phase-only and Phase+MOLLI methods (P=0.024) while no difference was observed between MOLLI and EPI-IR AIF_w values (P=0.096). AIF values computed with the two methods are shown in Fig. 4, overlaid with the gold-standard EPI-IR results.

Discussion

Despite widespread clinical application of DCE-MRI, it suffers from a lack of standardization and accuracy, especially with respect to the AIF⁴. It is not possible to accurately measure high, rapidly changing [Gd] in the AIF from T₁ measurements alone as MR signal saturates at these concentrations. The use of phase information is more desirable where no such saturation occurs⁴. In the steady-state washout of the phase-derived AIF, inaccuracies may occur due to inadequate phase accrual, flip angle error, partial-volume effects, and high temporal resolution requirement^4,5. The Phase+MOLLI method presented is accurate and fast for computing the entire AIF for DCE-MRI (Fig. 4). MOLLI T₁ measurements used to measure AIF_w were validated in static phantoms with gold-standard T₁ measurements (EPI-IR) (Fig. 3). Flow was found to have a significant effect on MOLLI 1/T₁ values between pre- and post-DCE (Fig. 1). [Gd] accuracy from MOLLI can be improved by using a calibration curve generated from a combination of MOLLI-Brain and MOLLI-Blood measurement parameters by using flow-insensitive MOLLI-Combination (Fig. 2).

Conclusion

There is currently no gold-standard for determining the AIF in DCE-MRI. The Phase+MOLLI technique provides an important improvement over previous methods since it avoids signal saturation and gives correct [Gd] values for the washout of the AIF. The Phase+MOLLI method provides an accurate, reproducible, and flow-insensitive AIF for clinical DCE-MRI.

Acknowledgements

The authors would like to thank the Ontario Research Fund and GE Healthcare who provided funding for the phantom used in our experiments.