Fast Multi-Parametric Mapping Competition: MR Fingerprinting vs. Triple-Echo Steady State
Christian Guenthner1, Thomas Amthor2, Sebastian Kozerke1, and Mariya Doneva2

1Institute for Biomedical Engineering, ETH Zurich, Zurich, Switzerland, 2Philips Research Europe, Hamburg, Germany


Magnetic Resonance Fingerprinting (MRF) and triple-echo steady-state (TESS) are two sequences that both allow for the simultaneous quantification of T1 and T2. While MRF relies on the transient response of tissue and noise-like under-sampling artifacts, TESS acquires the two lowest order SSFP-FIDs and the lowest order SSFP-Echo in the steady-state of a rapid, spoiled SSFP sequence.

In this work, we compare the performance of the two sequences in a phantom study, where imaging parameters and total acquisition duration between the two scan techniques were matched. In addition, a slice-profile correction for TESS is proposed and included in the comparsion.


Magnetic Resonance Fingerprinting (MRF) allows for the simultaneous quantification of tissue parameters by matching the transient response to a series of varying flip-angles (FA) and repetition times (TR) to a pre-calculated dictionary.1,2 Another technique for rapid relaxometry is the triple-echo steady state (TESS) sequence, which allows for the simultaneous estimation of T1 and T2 through acquisition of the two lowest order SSFP-FID$$$\;F_{+1,0}\;$$$and the lowest order SSFP-Echo$$$\;F_{-1}\;$$$in a rapid SSFP sequence with constant TR and FA.3 While MRF heaviliy relies on spatio-temporal correlations and noise-like undersampling artifacts, TESS requires the acquisition of consistent, high SNR images to allow fitting of a fully analytic signal model.

In this work, we propose a slice profile correction for TESS and compare the performance of MRF and TESS in a phantom study, where imaging parameters and total acquisition duration between the two scan techniques were matched.


Cartesian TESS3 (Fig.1) and spiral FISP-MRF2 (Fig.2) were implemented on a 3T Philips Ingenia system (Philips, Best, The Netherlands) and data was acquired using an 8-channel head coil. The measurements were performed on a phantom (High Precision Devices, Inc., Model 130) with 14 samples of differing T1 and T2 values. Both sequences were set up with equal imaging parameters: 240mm square FOV, 1.5mm isotropic resolution, single-slice excitation, 5mm slice thickness, spoiling in through-plane direction of ΔkS=4·2π/5mm and in the case of TESS also with additional in-plane spoiling of Δkx=2π/1.5mm spoiling area in both measurement and phase-encode direction (Fig.1b), and identical sinc-Gauss excitation pulse. The slice profile was determined using the hardpulse approximation. In addition,$$$\;B_1^+(\vec{r})\;$$$ was determined using actual flip angle imaging.4

TESS: The three spoiled magnetization states were acquired in successive TR periods by adding a respective re-phasing gradient before the readout and ensuring equal total spoiling in each TR (Fig.1).5 Sequence timing is depicted as part of Figure 1. Each echo was sampled three times to increase SNR and to match the acquisition duration of TESS and the MRF scan to 16s.

T1/T2 were simultaneously recovered using a least-squares minimization of the signal ratios$$S_{T1}\left(T_1,T_2\right)=\frac{F_{-1}}{F_0-F_1}\quad\text{and}\quad{}S_{T_2}\left(T_1,T_2\right)=\frac{F_{1}}{F_0}$$using an analytic signal model.6 The local$$$\;B_1^+\;$$$was included by per-pixel calculation of the effective FA. Moreover, a slice profile correction (SPC) for TESS was devised, which takes the finite spoiling$$$\;\Delta{}k_S\;$$$over the slice of thickness$$$\;\Delta{}s\;$$$into account. The corrected echo amplitudes are given by$$F_n^\text{SPC}=\sum_{k,z}F_+\left(z;k\right)e^{-\text{i}z\cdot\Delta{}k_S\cdot\left(k-n\right)}\cdot\text{sinc}\left(\frac{\Delta{}z\Delta{}k_S\cdot\left(k-n\right)}{2}\right),$$where$$$\;n\;$$$is the echo order ($$$n=-1\dots{}1$$$),$$$\;k\;$$$the contributing configuration states (here$$$\;k=-25\dots{}25\;$$$) and$$$\;z\;$$$the position within the discretized slice with bin width$$$\;\Delta{}z$$$. The configuration states$$$\;F_+\left(z;k\right)\;$$$were evaluated for each bin separately using the FA given by the product of local slice profile, B1+, and nominal FA.

Additional in-plane spoiling simplifies the slice profile correction to averaging over the profile according to$$F_n^\text{SPC, in-plane}=\sum_{z}F_+\left(z;n\right).$$

MR Fingerprinting: The MRF signal evolution was acquired using a single shot, 15-fold undersampled spiral acquisition with conjugate phase reconstruction to account for B0 inhomogeneity. The acquisition of 1000 time-points was preceded by an adiabatic inversion pulse with 20ms delay. A constant TR of 16ms and a varying FA up to 70° was used (Fig.2).

Parameter maps were obtained by matching of the temporal response to a pre-calculated dictionary using the EPG formulation.7,8,9 Two dictionaries were created, one only accounting for T1/T2 and a second dictionary accounting for B1+ and slice-profile determined prior to matching. Details are given in Fig.2c-d.

Results and Discussion

Fig.3 shows the dependency on T1/T2 of the TESS signal ratios$$$\;S_{T_1}\;$$$and$$$\;S_{T_2}\;$$$without and with SPC for in-plane and additional through-plane spoiling. TESS without SPC leads to underestimation of T1.$$$\;S_{T_1}\;$$$is equivalent for all models below a T2 of approx. 75ms, thus small T2 are expected to be unaffected by slice profile effects.

Fig.4 shows mean and standard deviation of T1/T2 values obtained with MRF and TESS against vendor-provided ground-truth (GT) values. MRF allows to recover T1 over the whole investigated range and underestimates T2 especially for values below 50ms. TESS without SPC underestimates T1 by a factor 2, which is in accordance with the findings of Fig.3. TESS with SPC and through-plane spoiling, allows to recover T1 up to 1250ms, however, minimization becomes unstable at low T1 values. Independent of SPC, TESS is in excellent agreement with GT for T2 values below 50ms. The discrepancy of TESS with additional in-plane spoiling compared to through-plane spoiling only, indicates that the simplified SPC model is not accounting for all slice-profile effects.


While MRF allows to recover T1/T2 estimates over the whole investigated domain, TESS shows excellent performance for short T2 estimation. T1 estimation with TESS highly benefits from SPC, however, convergence issues for low T1 values have still to be addressed. SPC for MRF does not improve the T2 estimation, thus further investigation of additional confounding factors is needed.


This project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 668039.


  1. Ma, D. et al. Magnetic resonance fingerprinting. Nature 495, 187–192, DOI:10.1038/nature11971 (2013).
  2. Jiang, Y., Ma, D., Seiberlich, N., Gulani, V. & Griswold, M. a. MR fingerprinting using fast imaging with steadystate precession (FISP) with spiral readout. Magn. Reson. Medicine 74, 1621–1631, DOI:10.1002/mrm.25559 (2015).
  3. Heule, R., Ganter, C. & Bieri, O. Triple echo steady-state (TESS) relaxometry. Magn. Reson. Medicine 71,230–237, DOI:10.1002/mrm.24659 (2014).
  4. Yarnykh, V. L. Actual flip-angle imaging in the pulsed steady state: A method for rapid three-dimensional mapping of the transmitted radiofrequency field. Magn. Reson. Medicine 57, 192–200, DOI:10.1002/mrm.21120 (2007).
  5. Heule, R., Celicanin, Z., Kozerke, S. & Bieri, O. Simultaneous multislice triple-echo steady-state (SMS-TESS) T1, T2, PD, and off-resonance mapping in the human brain. Magn. Reson. Medicine 80, 1–13, DOI:10.1002/mrm.27126 (2018).
  6. Hänicke, W. & Vogel, H. U. An analytical solution for the SSFP signal in MRI. Magn. Reson. Medicine 49,771–775, DOI:10.1002/mrm.10410 (2003).
  7. Scheffler, K. A pictorial description of steady-states in rapid magnetic resonance imaging. Concepts Magn. Reson. 11, 291–304, DOI:10.1002/(SICI)1099-0534(1999)11:5<291::AID-CMR2>3.0.CO;2-J (1999).
  8. Weigel, M. Extended phase graphs: Dephasing, RF pulses, and echoes - pure and simple. J. Magn. Reson. Imaging 41, 266–295, DOI:10.1002/jmri.24619 (2015).
  9. Malik, S. J., Padormo, F., Price, A. N. & Hajnal, J. V. Spatially resolved extended phase graphs: Modeling and design of multipulse sequences with parallel transmission. Magn. Reson. Medicine 68, 1481–1494, DOI:10.1002/mrm.24153 (2012).


Figure 1: Sequence diagram of the triple-echo steady state (TESS) sequence used. The three echoes F0, F-1, and F+1 and three averages are read out in consecutive sequence repetitions of equal phase encoding. Single slice excitation with slice thickness Δs is performed with spoiling in two variants. a) Through-plane spoiling only with 4·2π/Δs. b) Additional in-plane spoiling with 2π/Δx in measurement and phase encode direction. Δx denotes the isotropic in-plane resolution.

Figure 2: a) Sequence diagram of the employed MR Fingerprinting sequence based on spoiled SSFP with constant TR and TE.2 1000 undersampled images are acquired using an Archimedian spiral readout. The flip angle train is given in b). Acquisition, reconstruction and dictionary details are given in c-e).

Figure 3: a) Relative flip angle and phase profile of the used sinc-Gauss pulse for the TESS and MRF acquisitions. b) Depiction of the ST1 and ST2 ratios as defined by Heule et al. with and without slice profile correction for cases with through-plane spoiling only (yellow) and with additional in-plane spoiling (red). The configuration states were calculated according to the analytical formulas of Hänicke.6 From the signal dependency of ST2 at fixed T2 (top left plot), it is evident that TESS reconstruction without slice profile correction will underestimate T1. For T2<75ms, the ST1 signal is independent of the slice profile correction employed.

Figure 4: Results of MRF vs. TESS comparing T1 and T2 values against vendor provided ground truth (GT). MRF with and without SPC are approximately equal. TESS without SPC highly underestimates T1 as already observed in Fig.3. The SPC with through-plane spoiling allows to recover T1 up to 1250ms, however, minimization becomes unstable at low T1 values. T2 estimated by TESS is in excellent agreement with GT for values below 50ms, while MRF highly underestimates small T2. Generally, MRF allows to recover T1 and approximate T2 values over the whole investigated domain, whereas TESS shows high accuracy especially for low T2 values.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)