Introduction

MR imaging based on endogenous chemical exchange mechanism, such as Chemical Exchange Saturation Transfer (CEST), can be used to detect metabolites and proteins level in tissues. It is reported that R1rho (1/T1rho) asymmetry imaging may have certain advantages compared to CEST MTR asymmetry.^1,2 However, the conventional spin-lock approaches used to obtain R1rho are susceptible to B1 RF and B0 field inhomogeneities. Recently, an approach termed AC-iTIP was proposed to address this problem.³ However, an accurate B0 map may still be needed to retrospectively calculate R1rho asymmetry from AC-iTIP data. This is illustrated in Figure 1. In this work, we report a method to obtain tissue properties from AC-iTIP by fitting the entire R1rho-spectrum to the existing MRI physical models instead of performing R1rho asymmetry.

Methods

R1rho at different frequency offset (FO) can be represented by a superposition of three terms, namely relaxation due to the water pool ($$${R}_{{eff}}$$$), relaxation due to the chemical exchange ($$${R}_{{ex}}$$$) and relaxation due to the magnetization transfer ($$${R}_{{MT}}$$$): $${R}_{1\rho}({FO})={R}_{{eff}}+{R}_{{ex}}+{R}_{{MT}}, [1]$$ where $$${R}_{{eff}}={R}_{1}\cdot {cos}^{2}\theta+{R}_{2}\cdot {sin}^{2}\theta$$$. Trott and Palmer⁴ and Zaiss et. al⁵ have derived two equations using different assumptions. In this work, we use Trott and Palmer’s expression for its simplicity and focuses on fast chemical exchange. $${R}_{{ex}}=\frac{{p}_{{b}}\cdot {k}\cdot \Delta {\omega}^{2}}{(\Delta {\omega}-{FO}^{2})+{w}_{1}^{2}+{k}^{2}}, [2]$$ where pb denotes the population ratio of the labile proton to the water proton, k denotes the CE rate, $$$\Delta \omega$$$ denotes the chemical shift of the metabolite pool and $$$w_{1}=2\pi \cdot FSL$$$.

Zaiss et. al⁵ have derived an expression for $$${R}_{{MT}}$$$: $${R}_{{MT}}=\frac{(\Delta {\omega}^{2}+{r}_{2{a}}^{2})({k}_{{ba}}{r}_{1{a}}+{r}_{1{c}}({k}_{{ab}}+{r}_{1{a}}))+{w}_{1}^{2}{r}_{2{a}}({k}_{{ba}}+{r}_{1{c}})}{(\Delta {\omega}^{2}+{r}_{2{a}}^{2})({k}_{{ab}}+{k}_{{ba}}+{r}_{1{a}}+{r}_{1{c}})+2{r}_{2{a}}({k}_{{ba}}{r}_{1{a}}+{r}_{1{c}}({k}_{{ab}}+{r}_{1{a}}))+{w}_{1}^{2}({r}_{2{a}}+{k}_{{ba}}+{r}_{1{c}})}, [3]$$ where $$$r_{1a}=R_{1}-R_{eff}$$$, $$$r_{2a}=R_{2}-R_{eff}$$$, $$$r_{1c}=R_{1}+R_{rfc}-R_{eff}$$$, and $$$R_{1}=1/T_{1}$$$, $$$R_{2}=1/T_{2}$$$, $$$\Delta\omega=2\pi \cdot FO$$$, and $$$k_{ba,ab}$$$ are the exchange rate of magnetization transfer. The saturation rate $$$R_{rfc}(\Delta\omega)=w_{1}^{2}{\pi}g(\Delta\omega)$$$, where $$$g(\Delta\omega)$$$ defines the absorption line-shape of different types of tissue⁶.

Fitting is performed using non-linear least square method. There are 10 unknows in the fitting Equation 1. The initial value ($$$A_{0}$$$), lower ($$$l_{b}$$$) and upper ($$$u_{b}$$$) boundary conditions are given in the table 1, where the initial values are obtained from literature⁷.

Datasets were collected from four X-ray diagnosed OA patients with grade 2 and 3 and one healthy volunteer under the approval of the institutional review board. The MRI exams were conducted on a Philips Achieva TX 3.0T system (Philips Healthcare, Best, the Netherlands) using receiver knee coils. AC-iTIP scan parameters include: FO from -250 to 250Hz with 25Hz increment, FOV 16 by 16 cm², slice thickness 5mm, echo train length 27, TR/TE 2000/7.4ms, and SPIR after spin-lock for fat suppression. A standard dual-echo gradient echo B0 map is collected, which is served as the initial B0 value for fitting. Acquisition of one slice of AC-iTIP takes 5 min 38 sec. We only collected 3 slices of knee joint near the locations with obvious OA symptom. ROIs, as shown in Figure 2, were drawn on cartilage, which was supervised by a radiologist with 7 years of experience in MSK MRI. We calculated the R1rho asymmetry and fitted pb value at each ROI.

Results and Discussion

Figure 3 and 4 show the comparison of R1rho asymmetry and pb value between the healthy volunteer and the OA patients. Note that R1rho asymmetry of healthy volunteer shows a similar trend compared to the previously reported results of AC-iTIP.³ Also note the lower signal in OA patients compared to the healthy volunteer, which likely due to the reduced GAG concentration in the cartilage of the OA patients.

Conclusion

We demonstrated that AC-iTIP is promising in detecting GAG content change in the cartilage at 3.0T.

Acknowledgements

This study is supported by a grant from the Innovation and Technology Commission of the Hong Kong SAR (Project MRP/001/18X), Faculty Innovation Award from the Faculty of Medicine, the Chinese University of Hong Kong, and a grant from the Research Grants Council of the Hong Kong SAR (Project SEG CUHK02).