Myocardial T₂ and Apparent Diffusion Coefficient (ADC) can be used to evaluate the presence of edema and myocardial infarction^1,2. We recently developed a technique for joint acquisition and reconstruction of T₂ and ADC maps using a spin-echo EPI diffusion weighted sequence (SE-EPI DWI) with a total acquisition time short enough to accommodate measurement in a single breath-hold. Joint T₂/ADC mapping has improved precision and accuracy compared to independently acquired and reconstructed T₂ and ADC maps with equivalent scan time.³ The objective of this study was to optimize the joint T₂/ADC acquisition to enhance sensitivity to myocardial infarction and to evaluate performance in simulations and phantom experiments.

A joint T₂/ADC protocol must acquire sufficient SE-EPI DWI over a range of echo times (TE) and diffusion encodings (b) to measure both parameters within an acceptable breath hold duration (≤18s). Reconstruction involves fitting these SE-EPI DWI to a bi-exponential signal model: S(TE,b)=S₀ exp⁡(-bD)exp⁡(-TE/T2). Because TE and b are both varied (TE₁,b₁,TE₂,b₂,…,TE_n,b_n), many different protocols are possible. To reduce this complexity, we employed a 3-point method for joint T₂/ADC measurement, which fixes the SE-EPI DWI acquisitions to only three combinations of b and TE: 1) b=0 and the shortest possible TE (TE_Min); 2) b=0 and a second TE=TE₂ where TE₂>TE_Min; and 3) b>0 and the minimum TE for this b-value (TE_D). The three images are acquired using N₁, N₂ and N_D repetitions during a scan with fixed total time (N₁+N₂+N_D=18). Therefore, in the 3-point method, the degrees of freedom to specify the acquisition scheme are: TE₂, b and N₁:N₂:N_D. This 3-point method is a natural generalization of the widely used 2-point method in SE or DWI acquisition optimization.⁴ Simulations: Bloch equation simulations were used to simulate signals acquired with any combination of TE and b for T₂=61ms,ADC=2.16∙10^-3 mm²/s (approximating infarcted myocardium). Complex Gaussian white noise was added independently to 1600 signals such that the signal to noise ratio (SNR) for the highest signal image (TE_Min=21ms, b=0) was 40. T₂ maps and ADC maps were reconstructed using linear least squares fitting of the simulated signals to the signal equation described above. Mapping accuracy and precision were determined by the bias and standard deviation (SD) of the reconstructed T₂ and ADC values compared to programmed values. Experiments: A 50ml homogeneous agar and CuSO₄ gel phantom (T₂=61ms, ADC=2.16∙10^-3 mm²/s ) was imaged on a 3.0 T scanner (Siemens Prisma) using a SE-EPI DWI sequence with sequence parameters TR=1000ms, 2.5x2.5x5mm resolution with a range of TE: 21-85ms and b: 100-900s/mm². The accuracy and precision of simultaneous estimates of the joint T₂ and ADC reconstruction were investigated using subsets of the acquired data and various schemes spanning a range of TE₂ (27-85ms) and b (100-900s/mm²), TE_D based on scanner limits(54-76ms), and a range of repetitions N₁:N₂:N_D =[1:2:15, 2:4:12, 3:6:9] for both simulated and experimentally acquired data.

Figure.1 shows the simulated precision of T₂ and ADC estimates as a function of TE₂ and b (N₁:N₂:N_D=2:4:12 fixed). Figure.2 shows the precision dependence of T₂ and ADC estimates on N₁:N₂:N_D and TE₂ (b=700s/mm² fixed) by simulations and experiments. Simulations show that high precision ADC estimates (SD_ADC<5%) require TE₂≥60ms and 400s/mm²≤b≤700 s/mm² (Fig.1). To achieve high precision T₂ estimates (SD_T2<5%), TE₂ must be ≥60ms. T₂ precision did not depend on b. N₁:N₂:N_D=3:6:9 was the optimal repetition ratio with maximum gains in ADC precision of about 5% and in T₂ precision of about 9%. (Fig.2) Phantom experiments were generally in agreement with simulations, with N₁:N₂:N_D =3:6:9 being optimal for most b-values and TE₂ (shown for b=700s/mm² in Fig. 2). We further validated the optimized acquisition through comparison with the non-optimized acquisition scheme. The results show general improvement in the accuracy and precision of T₂ and ADC maps (Table 1).

For a fixed tissue type, Bloch simulations were used to define the optimal acquisition (TE₁, TE₂, N₁:N₂:N_D) for joint T₂/ADC mapping. These parameters were then confirmed in phantom experiments. Note that while simulations set a maximum SNR=40 to be consistent with expected in vivo data, the phantom experiments had much higher SNR (SNR_Max=140). This likely explains the discrepancies between phantom experiments and simulations. Future work will involve implementing the optimized acquisition schemes in in-vivo joint T₂/ADC mapping in the heart.The accuracy and precision of both T₂ and ADC maps were improved for joint T₂/ADC mapping with an optimized, 3-point acquisition. Improvements were demonstrated in both simulations and phantom experiments.