Researchers in the fields of lung imaging and ultrashort echo-time (UTE) imaging.Motion due to respiration is one of the major difficulties in lung imaging. This problem is particularly pronounced in mice, which have a 10-20-fold higher respiratory rate than humans. To compensate for respiratory motion in vivo, MR acquisitions are usually synchronized with the respiration cycle either using a ventilator or prospective respiratory gating. In many cases spontaneous breathing is advantageous to avoid ventilator-induced lung injury or for physiological measurement. Further, prospective gating errors result from abnormal respiration waveforms caused by body motion. These problems can be avoided by retrospective gating (i.e., binning data after acquisition to avoid periods of motion). In traditional Cartesian imaging, retrospective gating typically requires that specialized sequences be employed. However, the situation is different in center-out radial encoding because respiratory motion slightly alters the local field experienced by spins within the lung, providing a detectable signal from the initial point of free induction decay (FID). Thus, the signal amplitude of each FID (i.e., k = 0) can be used for retrospective "self"-gating. Here we demonstrate that this retrospective "self"-gating can be implemented in mice using 3D UTE.All experiments were approved by Institutional Animal Care and Use Committee. Imaging was performed using a Bruker 7T scanner on 10 free-breathing 7-week old C57BL/6J mice. Mice were anesthetized with isoflurane and placed supine inside a quadrature birdcage coil (length: 50 mm, inner diameter: 35 mm). In addition, a pressure sensor of a small animal monitoring system (SA Instruments, Inc.) was taped on the mouse stomach to monitor the respiratory motion. Spherical k-space coverage was used for 3D UTE MRI with 2D golden means determining the azimuthal and polar angles of the endpoint of each radial spoke¹. The RF excitation pulse was applied with a slab selection gradient to focus the field-of-view (FOV) on the lung, with data acquisition starting immediately when the readout gradient was ramped after a slab refocusing gradient. A linear phase increment of 117° × RF pulse number added to each RF pulse was combined with a negating readout gradient and a strong spoiler gradient for spoiling². Total acquisition time was 12 minutes with TR = 7 ms, TE = 0.63 ms, FA = 5°, FOV = (30 mm)³, voxel = (0.23 mm)³, 64 points along each projection and 2-fold oversampling of projections. The respiration rate of one mouse was changed by adjusting isoflurane level to investigate the period of the FID signal amplitude as a function of projection number. FID signal amplitude as a function of projection number was smoothed by a moving-average filter and the resulting smoothed first- and second-order derivatives were combined to extract data at expiration and inspiration respectively, as shown in Fig. 2. The retrospective-gated radial data were resampled onto a Cartesian grid by convolving with a Kaiser-Bessel window³ before Fourier transform. Tidal volume was calculated by measuring the lung volume difference between the expiration and inspiration images. Lung parenchyma SNR was defined as the lung parenchyma signal divided by the standard deviation in the background.FID signal amplitude as a function of projection number at 3 different monitored respiration rates (~110/min, 80/min, and 60/min) is shown in Fig. 1, with plateaus and troughs corresponding to expiration and inspiration respectively. The respiration rates estimated from the amplitude curves are 109/min, 75/min and 64/min respectively, consistent with external monitoring. Fig. 2 shows the projections extracted for expiration (49462 projections) and inspiration (9864 projections) marked by the red and blue lines respectively. The resulting coronal and sagittal images are shown in Fig. 3. The red lines emphasize the diaphragm displacement between expiration and inspiration. Lung volume measured at inspiration and expiration is 0.55 ml and 0.44 ml respectively, resulting in 0.11 ml of tidal volume, consistent with previous plethysmographic measurements in (non-anesthetized) C57BL/6J mice of 0.16 ml (range 0.13-0.21 ml)⁴. The lung parenchyma SNR at inspiration (11.2) is lower than that at expiration (21.2) because of bulk density decrease, motion artifacts, and projection undersampling. The expiration/inspiration signal-intensity ratio in lung parenchyma is 1.2, consistent with the inspiration/expiration lung-volume ratio of 1.25, assuming similar T₂^*.We demonstrate the feasibility of retrospective-gated 3D UTE pulmonary MRI on small animals. The high parenchymal SNR in both expiration and inspiration images enables quantitative analysis of lung parenchyma and lung tidal volume, which match physiological expectations.