Heteronuclear proton MRI (also Highly Shifted Proton MRI) is a recent method for MRI cell tracking [1]. Similar to fluorine MRI, direct detection of distinct cellular populations can be realized using different Ln-DOTMA complexes to label cells. DOTMA provides a NMR resonance line which is highly shifted, depending on the lanthanide used, and can be separated from the water line (Fig. 1). The detected DOTMA signal has extremely short relaxation times [2], which allows for efficient signal amplification by paramagnetic relaxation enhancement, but also requires ultra-fast readout such as ultrashort echo time (UTE) to avoid relaxation losses. The method enables multicolor MRI, as resonance lines of different Ln-DOTMA complexes are separated (Fig. 1). The chemical shifts of these lines are sensitive against temperature changes [2], potentially causing problems for UTE acquisition. An alternative readout is Chemical Shift Imaging (CSI), also minimizing relaxation losses and enabling acquisition of different Ln-DOTMA complexes in one scan for efficient multicolor MRI. Yet, longer TR as compared to UTE reduces the SNR of CSI. Here, we compare UTE and CSI as readout for Heteronuclear Proton MRI in terms of SNR, sensitivity towards temperature changes and performance in separating resonance lines from different Ln-DOTMA complexes.MRI was performed at 9.4T at constant temperature (21°C ). Ln-DOTMA complexes were dissolved in pure water: Tm-DOTMA (Macrocyclics, Texas, USA) 10 mM and 50 mM, Tb-, and Dy-DOTMA (synthesized in-house) 10 mM. The DOTMA methyl resonance was observed at -45,+25,+30 kHz for Tm-,Tb-,Dy-DOTMA respectively, in accordance with [2].To assess the SNR 2D and 3D experiments for both, UTE and CSI, were performed with Gaussian excitation pulse with narrow bandwidth (EB: 2kHz) in order to excite only methyl resonances of Tm-DOTMA (offset: -45kHz, 21°C ) and not water, varying acquisition bandwidth (AB : 3.2–25kHz), minimum TR, Ernst angle excitation and 400x400µm³ (4-mm slice)/781x781x781µm³ resolution. Since the chemical shift changes due to temperature changes are known [2], we assessed the effects of temperature variation by keeping temperature constant and simulating chemical shift changes by gradually shifting 1) the frequency of the excitation pulse while leaving the acquisition window centered to Tm-DOTMA peak, and 2) the acquisition window with on-resonant excitation of Tm-DOTMA. AB was 25kHz. For multicolor MRI, UTE was performed with narrow excitation pulses (2kHz) to excite only one resonance. CSI used broad excitation pulses (6kHz) to excite both, Dy- and Tb-DOTMA.Minimum scan time was substantially different for CSI and UTE read out, 169s, versus 12s (for 20kHz AB) for a full 3D data set. CSI resulted in undistorted image without artifacts for all acquired AB . SNR decreased exponentially with larger AB. For UTE, SNR was strongly dependent on AB and reached a maximum at 20kHz, almost four times higher than CSI. For AB below 10kHz, blurring artifacts were observed in UTE images (Fig. 2). Since narrow excitation pulses were used for both UTE and CSI, temperature shifts may result in incomplete excitation. This effect was mimicked by shifting the excitation frequency at constant temperature. For both CSI and UTE, this resulted in a sensitivity profile reflecting the pulse shape, and for UTE resulting in 20% SNR loss for a shift corresponding to ± 3°C (Fig. 3A). For UTE temperature changes had a second effect: Shifting the resonance away from the center of the acquisition window, which we mimicked by shifting the acquisition window, resulted in increasing artifacts. A 20 % SNR loss was observed at a frequency shift corresponding to ±5°C (Fig. 3B) Frequency offset in the acquisition window had no effects on CSI. Multicolor MRI was possible with both readout schemes. Dy- and Tb-DOTMA could be separated in both CSI (Fig. 4) and UTE (Fig. 5). SNR of Dy-/Tb-DOTMA was 16.8/10.5 for UTE and 8.0/6.6 for CSI.For heteronuclear proton MRI, UTE readout provides faster image acquisition and higher SNR than CSI. Even for multicolor MRI measurements with different Ln-ions, two separate UTE acquisitions provide higher SNR than a single CSI measurement for Tm- and Dy-DOTMA. For Tb-DOTMA similar SNR efficiencies were observed. However, UTE is prone to temperature shift-induced artifacts, which may occur due to high duty cycle of the MR sequence. CSI on the other hand is insensitive to temperature changes because broad excitation pulses can be used, which do not compromise the ability to separate resonance lines in CSI. For future in vivo cell tracking measurements, however, temperature effects appear less important, since the body temperature of the subject counteracts potential MR-induced warming.