Recent advances in hyperpolarization (HP) ¹³C methods have allowed exceptional characterization of metabolism and perfusion changes, specifically with HP pyruvate and urea MRI, respectively, that underlie various diseases processes^1,2. The ability to obtain both sets of information and/or resolve multiple compounds in one scan via copolarization has been demonstrated in vivo in prior hyperpolarized ¹³C studies³. Several different k-space acquisitions have been utilized to achieve this spectral selectivity, including echo-planar spectroscopic imaging and echo-planar imaging^4,5, although these acquisitions typically have lower spatial resolution. The bSSFP sequence offers the highest SNR per unit time, and consequently highest spatial resolution, but exhibits off-resonance effects that make spectral selectivity relatively more difficult⁶. While different methods for spectral selectivity have been successfully applied to the bSSFP sequence, particularly at high field^7,8, adapting the sequence for high resolution 3D dynamic acquisitions at the clinically relevant field strength of 3T is more challenging. In this project we aimed to investigate two different methods for spectrally selective high resolution 3D dynamic bSSFP of copolarized [2-¹³C]pyruvate and [¹³C,¹⁵N₂]urea at 3T.Two different methods for spectrally selective bSSFP tested on a 3T GE MR scanner were (referred to by number from here on): (1) frequency encoding of both species simultaneously using a low gradient with wide readout bandwidth (“multiband” encoding)⁹; (2) individual acquisition of each species applying spectrally selective RF pulses with TR chosen to place the compound of interest in the passband and the other in the stopband. Figure 1 depicts each method as it would be used to acquire images of copolarized [2-13C]pyruvate and [¹³C,¹⁵N₂]urea. ¹³C thermal phantom tests were performed with a custom 3D bSSFP sequence (single timepoint) to display efficacy of each of the methods. Two spherical phantoms, one filled with bicarbonate and one filled with acetate, were imaged. Method 1 aimed to separate the two phantoms by low bandwidth frequency encoding, while method 2 aimed to alternately excite and suppress an individual spectral line (i.e. bicarbonate signal) when on-resonance and off-resonance, respectively. In vivo experiments were initially performed on one Sprague-Dawley rat with copolarized [2-¹³C]pyruvate and [¹³C,¹⁵N₂]urea, which have a frequency difference of 1412 Hz at 3T. Both methods were acquired with a flip angle of 50°, after a 25° preparatory pulse (no slice select gradients), 2.5 mm isotopic resolution, 6 s temporal resolution for each compound, 10 dynamic timepoints for a total scan time of 60 s, and the scans were started at 15 s after beginning of injection. The different experimental parameters were as follows: (1) 48 x 24 x 8 matrix size, 1.6 ms pulse with a TBW of 4, TR of 21.9 ms, and rBW of 2.824 kHZ; (2) 24 x 24 x 8 matrix size, 2 ms pulse with a TBW of 2, TR of 7 ms, and rBW of 15.625 kHZ. Both phantom and in vivo experiments were acquired using partial Fourier acquisition, with only ¾ of the phase encodes acquired. A projection onto convex sets (POCS) reconstruction was employed to reconstruct the raw data, and further reconstruction for the multiband data was performed as described previously⁹. In vivo DNP experiments used a HyperSense polarizer, and 3mL of copolarized 80 mM [2-¹³C]pyruvate/80 mM [¹³C,¹⁵N₂]urea was injected via a tail vein catheter.The phantom experiments successfully showed the capability of using both methods at 3T. Figure 2 shows the representative slice ¹³C images for both methods. Method 1 successfully features bicarbonate and acetate in two separate bands, and both spherical phantoms are in their correct location in the recombined image. Method 2 successfully demonstrated that choosing the appropriate TR results in the bicarbonate signal being located in the stopband and consequently the spherical phantom disappearing. Figure 3 shows the 3D urea images next to a representative ¹H image of the rat for both methods for the first dynamic timepoint. Figures 4 and 5 show all 10 dynamic timepoints for both methods for urea and pyruvate, respectively. Method 2 appeared to have better SNR and longer lasting dynamics. The ability to acquire spectrally selective high resolution 3D dynamic images at 3T via the bSSFP sequence is demonstrated here using two separate methods, multiband encoding and spectrally selective RF pulses with an appropriate TR. Individual images of [2-¹³C]pyruvate and [¹³C,¹⁵N₂]urea were successfully acquired within one scan. These methods can be potentially applied to other systems that require imaging of multiple resonances, such as pH measurement with HP [¹³C]bicarbonate. With further optimization of each method, higher resolution can be achieved and applied to various disease processes, and ultimately translated to clinical imaging.