Researchers interested in imaging of microstructures in living tissue as well as other porous materials by means of nuclear magnetic resonance diffusion experiments.The study of porous microstructures is of high interest in medical imaging. The structure of the observed tissue on the scale of cell sizes can reveal important information about possible pathologies. While classical diffusion weighted imaging is only indirectly related to structural features like the cellularity, diffusion pore imaging (DPI), which builds on a modified diffusion weighting,^1,2 has recently been proposed as a means to acquire images of the average cell shape in a voxel or region of interest. In this work, we present the feasibility of DPI on a preclinical scanner for the first time. Further, preliminary results are shown for a comparison of two different DPI implementations as proposed by Laun et al.¹ and Hertel et al.³In contrast to the classical pulsed gradient spin echo (PGSE) sequence,⁴ which uses two narrow gradient pulses (Fig. 1a), the mentioned DPI technique is based on a highly asymmetric gradient shape with a long pulse of low amplitude and a narrow pulse of high amplitude. Due to this change, DPI is able to preserve the phase information of the scatter-like diffusion process, which fails in the classical approach. In the following, the implementation using a single long and a single narrow gradient pulse as proposed by Laun et al.¹ is labeled method 1 (Fig. 1b) and the implementation as used by Hertel et al.³ is labeled method 2 (Fig. 1c). The latter splits up the gradient pulses into multiple shorter ones and introduces accordingly as many additional refocusing pulses in between them. Both techniques were implemented on a preclinical 9.4T animal scanner (Bruker Biospec) with a maximum gradient amplitude of 706 mT/m. An ensemble of about 1000 water-filled glass capillaries with inner diameter of 20 μm was measured. Effective durations of 200 ms for the long and 8 ms for the narrow gradient pulse were the same in the two implementations. For method 2, the long gradient pulse was subdivided into 40 and the narrow one into two pulses.Figure 2 shows the real part of the 1D signal plotted over the applied q-values. A first order Bessel function J₁ was fitted to the data, which is the expected characteristic signal shape for the cylindrical pore shape, yielding fitted diameters of (20.6 ± 0.4) µm (method 1, top) and (19.3 ± 1.2) µm (method 2, bottom). The agreement of the signal with J₁ and the consequent presence of negative values indicates the achieved phase preservation. The SNR was approximately 51 for method 1 and 16 for method 2. Figure 3 shows the result of a 2D measurement with method 1 after a 2D-Fourier-transformation into position space and displays the average pore shape of the phantom. Remarkably, the resolution of this image is of the order of a few μm.Feasibility of DPI on a preclinical animal scanner was demonstrated for the first time. Method 1 and 2 were both well suited. We currently deem the reported difference in SNR values to be a preliminary result, which may change upon implementation of improved RF-amplitude adjustments. Method 2 results in a potential SNR increase due to shorter echo times. However, this comes at the cost of introducing many additional refocusing pulses along with a complex phase cycling scheme to suppress unwanted coherence paths. Method 1 builds on a single refocused spin echo sequence with one clear coherence path and the effect of imperfectly adjusted RF-pulses is minimized. For an in-vivo application of DPI, various practical issues, such as SNR, and also fundamental problems, like finding an appropriate way of dealing with extracellular water, will have to be tackled. However, the shown feasibility of phantom measurements on a preclinical animal scanner opens the possibility of a potential in-vivo measurement realization.