PT  - JOURNAL ARTICLE
AU  - Pretorius, P. Hendrik
AU  - Pijanowski, Justin
AU  - Segars, William
AU  - Cao, Xinhua
AU  - Chafe, Katelyn
AU  - Pells, Sophia
AU  - Kwatra, Neha
AU  - Treves, S. Ted
AU  - Fahey, Frederic
AU  - Yang, Yongyi
AU  - King, Michael
TI  - <strong>Separate respiratory motion estimation and compensation of the right and left kidney in pediatric DMSA SPECT imaging: anthropomorphic phantom simulations and clinical studies</strong>
DP  - 2024 Jun 01
TA  - Journal of Nuclear Medicine
PG  - 242261--242261
VI  - 65
IP  - supplement 2
4099  - http://jnm.snmjournals.org/content/65/supplement_2/242261.short
4100  - http://jnm.snmjournals.org/content/65/supplement_2/242261.full
SO  - J Nucl Med2024 Jun 01; 65
AB  - 242261 Introduction: It is known that the left and right kidney can have different respiratory motion amplitudes due to the position of the right kidney below the liver. Previously we introduced a processing strategy using three regions-of-interests (ROIs) defined around the kidneys to potentially isolate any differences in motion during the reconstruction process. The aim of this study was to determine whether it is possible to separately estimate and compensate for respiratory motion of the left and right kidney in anthropomorphic phantom simulations and clinical studies.Methods: A subset of previously generated XCAT phantoms with the respiratory motion of the left kidney set to 75% of the right kidney value were used. The SIMIND Monte Carlo package was used to simulate 100 msec sub-projections with acquisition parameters matching those of clinical studies. Similarly, pediatric patient data were acquired on a dual detector Siemens (Hoffman Estates, IL) gamma camera employing low-energy ultra-high resolution (LEUHR) parallel-hole collimators with a pixel size of 0.239 cm in list-mode which was framed into 100 msec time-bins. An initial reconstruction without respiratory motion compensation is used to define projections specific ROI’s for the left kidney and right kidney, and an ROI combining the ROIs of the left and right kidneys. The latter was used in determining the center-of-mass surrogate respiratory signals used for reframing patient data into 7 respiratory bins. The XCAT phantom 100 msec sub-projection sets were also reframed into 7 respiratory bins employing the true signal used to create the original respiratory motion during simulation for comparison. Three iterations of Maximum Likelihood Expectation Maximization (MLEM) reconstruction were used to reconstruct both phantom and patient data for the 7 respiratory bins. These were then used in respiratory motion estimation by registration between the 6 outer bins and the center bin. A reconstruction combining the data of the 1st and 2nd 180-degree detector rotation without motion correction was performed for usage in comparison to motion correction. Separate respiratory estimates are obtained for the left and right kidney by employing kidney specific ROIs. We deviated from our previous ROI reconstruction implementation in that only two projection-specific ROIs were used for the final reconstruction correcting for respiratory motion. The two ROI’s separate the left and right kidney and divide the background equally between them alleviating sharp edge-artifacts that were be visible with the 3-ROI implementation. MLEM with 20 iterations were used. The differences in respiratory estimates of the left and right kidney of phantom data were calculated and visual inspection of respiratory motion compensation without and with ROIs was performed.Results: XCAT phantom respiratory estimated amplitudes in the axial direction (head-to-feet) of the left kidney varied between 56.7% and 79.0% (average = 72.3%, standard deviation = 6.2%) when expressed as a percentage of the right kidney results in comparison to the simulated value of 75%. Patient respiratory estimates for the right and left kidney varied between 34.2% and 124.1% (average=85.5% standard deviation=26.6%). Inspection of the respiratory corrected slices for XCAT phantom patients using the ROI strategy seems to yield slightly better kidney visualization than without, when the respiratory amplitude is large (~24 mm).Conclusions: For the XCAT phantom, we determined a statistically significant difference in the left and right kidney respiratory amplitude in agreement with that simulated (p = 7.56E-0.5). In patients with unknown respiratory motion differences between kidneys, such differences were quantitatively larger but more difficult to visualize. However, we did note generally a visual improvement in the kidneys with motion correction.