Adenosine stress STEAM-IVIM: Optimised protocols and preliminary results (RF_FR_391)

Intravoxel incoherent motion (IVIM) imaging is a variant of diffusion tensor CMR (DT-CMR) that delivers simultaneous, contrast agent free noninvasive assessment of myocardial microstructure and microvascular perfusion. We recently demonstrated the high sensitivity of the STEAM sequence to changes in perfusion¹ and developed a STEAM variant delivering cardiac IVIM in-vivo at rest using a lengthy protocol².

To deliver clinically, IVIM must deliver measures interrogating the response of the myocardial blood supply to increased demand.  In this work we use simulations to develop an optimised shortened STEAM-IVIM protocol and demonstrate initial results with this protocol during adenosine stress.

Methods:

Simulations

STEAM-IVIM data acquired in 20 healthy volunteers (20 b-values, 2 averages²) was sub-sampled and compared to the full dataset to determine the optimal b-value protocol for a given number of breath holds. The diffusivity (D), perfusion fraction (f) and pseudo-diffusivity (D*) were calculated from data with 4-20 b-values. A boot strapping like analysis was used to determine the 4 b-values (1 average) resulting in the minimum normalised root mean square error for the combined parameters and the sequence of b-values that maximally reduces the error for each additional breath hold.

In-vivo imaging

Three healthy volunteers (2male, age 27-42 years) were imaged with breath hold STEAM IVIM at rest and during adenosine stress (140µg/kg/min) at 3T (Siemens Cima.X). During stress, STEAM IVIM data was acquired from the start until the end of the physiological response with ~5minutes of adenosine infusion, acquiring as many b-values as possible according to the order determined in the simulations.

 

Results:

Figure 1 plots the errors (bias and limits of agreement) in f, D and D* with increasing numbers of b-values for the optimal b-value acquisition order (0, 30, 300, 1000, 70, 100 150, 20, 250, 90, 600, 400, 10, 50, 40, 800, 500, 60, 80 and 200 smm^-2), showing reducing uncertainty and inaccuracy with an increasing amount of data included in the fitting process.

All 3 subjects successfully completed the stress STEAM IVIM scan with 10, 10 and 13 breath holds completed during adenosine. Figure 2 shows example D, and f bullseye plots at rest and stress in one example subject, demonstrating the expected increase in f during stress.

Figure 2 plots values of D and f from the three subjects at stress with reference to the normal ranges established in healthy volunteers using the full 20 b-value protocol in prior work², demonstrating the expected increase in f due to stress.

Conclusion:

We provide an optimised b-value protocol for shortened STEAM-IVIM. Breath hold STEAM-IVIM can be acquired during adenosine stress and demonstrates the expected increases in the perfusion fraction (f) that would be expected with vasodilation. Future studies will assess the performance of non-contrast stress STEAM-IVIM relative to gadolinium contrast based quantitative first pass perfusion imaging in patient cohorts.