Left ventricular blood TAC quantitation with microPET imaging in mice using MAP, FBP and blood sampling

Improved resolution with iterative maximum a posteriori (MAP) image reconstruction is promising; however in the past, image values varied based on the detected total counts, therefore filtered backprojection (FBP) has remained the first choice for quantitation using microPET. To determine if MAP can be utilized to generate accurate noninvasive blood and organ time activity curves (TACs) in mice using FDG, we compared ordinary Poisson MAP (OP-MAP) and MAP to FBP and arterial blood samples. C57BL/6 mice (n=13) were imaged for up to 90 min using a Focus 220 microPET. Blood samples (n=9 to 25, 5-15 /spl mu/L) from a femoral artery catheter were drawn after injection of 30-50 /spl mu/L FDG (17-37 MBq) and converted to PET counts. Data was binned into 32 frames and reconstructed using FBP (all 32 frames), MAP and OP-MAP (6 or more frames). Frame durations ranged from 0.3 sec to 15 min. Volumes of interest (VOI) were assigned to the left ventricular blood pool (LV, /spl sim/1.2 mm), whole body (WB), liver, whole heart and myocardium (average of the 4 walls). No correction was made for partial volume or myocardium spillover of activity. WB VOI values from the older version of MAP were not well matched to the FBP WB values and were variable depending on frame duration. OP-MAP VOI values were not substantially different from FBP values in WB and liver (no partial volume effect; PVE). LV values were higher during the first passage of the tracer bolus with OP-MAP, due to less PVE, and lower at later times due to less spillover of activity from the myocardium. Initial results indicate that OP-MAP image values are well matched to FBP values where PVE is not a factor, and that OP-MAP data are not count rate or time duration dependant. Further work is underway to verify that OP-MAP data can be useful for quantitative kinetic modeling.