T2/T2* based relaxometry is increasingly used to non-invasively quantify tissue iron content in lieu of biopsy. However, studies have shown that the observed linear relationship between R2 (1/T2) and tissue iron concentrations does not hold true above an iron concentration threshold (dubbed as the ‘saturation threshold’) [1,2]. Our numerical simulations and phantom experiments show that the choice of interval between the echo times used to sample T2 decay, and noise levels play a crucial role in determining the saturation threshold, and that the linear relationship between iron concentration and R2 can be extended by judiciously varying echo spacing.
1. Pierre et al, Noninvasive measurement and imaging of liver iron concentrations using proton magnetic resonance, BLOOD, 15 JANUARY 2005, VOLUME 105, NUMBER 2
2. Brown GC et al, Performance and limitations of R2* relaxometry liver iron measurements, December 2010, Proc. Intl. Soc. Mag. Reson. Med. 19At: Montreal
Fig 1: Surface plot of effects of true R2 value and noise on R2fit (1A) and its error (1B) with dTE 3ms. The Fig. 1A shows that linear relationship between R2 values and means R2fit at low noise and low R2. It becomes saturated at high R2 and high noise. The fig. 1B shows that error is high at very low R2 and then becomes flat for a short range and then increases again.
Fig 2: relationships of R2fit with true R2 with error bar of two standard deviations from mean R2fit: 2A) noise of 1%, 2B) noise of 10%, and top row is for short echo spacing (SES) with the dTE of 3 ms , middle row is for long echo spacing with the dTE of 12 ms, bottom row comparison between SES and LES. Compared to 1% noise, 11% noise significantly reduces linear range and becomes saturated quickly. The R2fit noise propagated from signal noise jumps when R2fit begins saturated.
Fig 3: R2 measurement on phantoms with a series of MnCl2 concentrations. Linear curve fitting was performed on a subset of data (MnCl2 from 0 to 7 mM) with dTE of 0.2 ms. This linear function was used to extrapolate R2 for the MnCl2 concentration beyond the linear range. The saturation occurs at high MnCl2 concentration and high dTE the same to our simulation. The equivalent liver iron concentration (LIC) was calculated according to Pierre’s paper (ref 1) and added to x-axis at the top.
Fig 4: The image intensity varies with MnCl2 concentration. Left top: initial image intensities of phantoms at the TE0; left bottom: decay factor exp(R2*dTE) at the second TE1 for different dTEs (legend on the right) where R2 values were computed using the fitting equation R2 = 71.23*MnCl2 + 8.99. Right: multiples of initial image intensities and decay factor at the second echo TE1. For the longest dTE of 3 ms, the signal at TE2 drops to 8.21% at MnCl2 of 4 mM and drops to 4.1% at MnCl2 of 5 mM.
Fig 5: The image intensity varies with MnCl2 concentration. Left top: initial image intensities of phantoms at the first TE1; left bottom: decay factor exp(R2*dTE) at the fifth echo TE4 for different dTEs (legend on the right) where R2 values were computed using the fitting equation R2 = 71.23*MnCl2 + 8.99. Right: multiples of initial image intensities and decay factor at TE4. For the shortest dTE of 0.2 ms, the signal at TE4 drops to 9.14% at MnCl2 of 5 mM and drops to 3.71% at MnCl2 of 7 mM.