Am I making valid correlations??

[ajm8957]

basic experimental design:

started with 60 mice at 6 weeks of age, using MRI and PET imaging, I defined the volume of a particular organ that changes as the animal ages- I also determined mean glucose uptake within that organ.

20 animals were sacrificed immediately after imaging- the organ was removed and weighed, and several other cell based function measurements were performed.

remaining 40 animals were aged to 24 weeks- all animals imaged (MRI & PET)- 20 animals sacrificed, organ removed- same assays performed as above.

Remaining 20 animals were aged to 52 weeks- all animals were imaged (MRI & PET)- animals were sacrificed, organ removed- same assays performed as above.

what I am trying to do is establish a correlation between the imaging data and the cellular measurements data- I have done a regression analysis- x axis is total number of cells in the organ for 60 animals (20 at 6 weeks, 20 at 25 weeks and 20 at 52 weeks); y axis is the PET/MRI measured glucose metabolism for that organ at that timepoint just prior to sacrifice (20 at 6 weeks, 20 at 25 weeks and 20 at 52 weeks). I have then done a Pearson correlation and am showing p value and r^2.

Cell numbers within this organ are known to decline with age, I am simply trying to establish an imaging approach as a viable method for non-invasively assessing the organ. I have done the same regression to correlate glucose uptake with other parameters of organ function.

Is this a valid approach. Someone told me I need to also do an ANOVA- but I am not sure if I need to.... PLEASE HELP!!??

[kingkong]

The correlation analysis is more appropriate, because you are just trying to establish a relationship between the imaging data and the cellular measurements data. Both variables are not in causation relationship. In correlation analysis, you don't need to define which one affect which one. But for ANOVA, you have to define which is the dependant variable and which is independent variable. In fact, there is no independent variable at your study.