Purpose: The universal pipeline for T1 mapping is to acquire series of images with different T1 weightings and fitting the signal intensity to T1 relaxation for each pixel location. Optimization of T1 relaxometry is mostly focused on the efficiency of T1 estimation while trading off the total length of the experiment and T1 mapping computation. This research investigates the number of TIs to optimize the precision of T1 measurements using the classical inversion recovery (IR) technique.
Methods: The standardized ACR phantom was scanned using IR sequences with a constant TR = 6000 ms and a range of TIs. A Matlab script was developed for the T1 mapping calculation based on the Barral JK methodology. Parametric maps were generated for 11 series of images each with different numbers of TIs (ranging from 3-6), and different choices of TI. The T1 map generated with 15TIs was considered to be the reference T1 map. The MultiScale_ Structural Similarity Index (MS_SSIM) was the chosen method for image analysis. The SSIM was computed between the reference map and each other experimental maps. One-Way ANOVA and Tukey comparison was used for statistical analysis.
Results: The p-value is less than 0.05 and the mean differences between the MS_SSIM of the different TIs is statistically significant. Also, the difference between the mean of 3TI vs. 4TI and 3TI vs. 5TI is statistically significant. The high predicted R2 value equal to 91.5% indicates that the model generates precise predictions for new observations. The groups of T1 maps generated with 3TIs showed statistically significant similarity to the reference map comparing with maps generated with 15TIs.
Conclusion: T1 mapping generated using 3TIs provide comparable results to the T1 mapping generated using 15TIs. This reduces scanner time for experiments.