I'm currently working with pesudo-dynamic PET scans, i.e. data containing sequentially acquired time frames. Using all the time frames I can 'easily' compute the Ki-values (slopes of the regression line according to http://doc.pmod.com/pkin/pmclass.lib.pmkin.models.PKpatlakNonInvasive.htm)
I wonder if results were still is valid if I did not have PET-data from the time directly after application of the tracer but only starting from a time t >>0. This would of course change the values of the integrals in the formula, but considering the linear model that is applied, the slope of the regression line should stay the same while only the intercept changes, right?
Thanks for your help!
http://doc.pmod.com/pkin/pmclass.lib.pmkin.models.PKpatlakNonInvasive.htm
Thanks, Benjamin, for your idea to simply try it. I have just compared two variants vor one sample voxel:
9 time frames (full acquisition time) vs. the last 6 time frames. Here you can find the Time-Activity-Curves for sample voxels and the reference region as well as the integrals over these values (trapezoidal numerical integration): https://www.dropbox.com/s/mrrwgvewul0e99c/Patlak_test_values.txt?dl=1
These are the resulting x-y-pairs: https://www.dropbox.com/s/j2oomnhw2w7hlp7/Patlak_test_new.png?dl=1
(red: 9 time frames, blue 6 time frames)
As you see, the slopes of the respective regression lines are not the same. (0.0772 vs. 0.0421) That's too bad.
So I would conclude that it is NOT POSSIBLE to calculate Ki-values if there is no continuous time-activity data available beginning from t_0 = 0 = application of radiopharmaceutical. I hope there is somebody out there who does NOT agree. :-)
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression