Question about ANOVA analysis

Dear Yan,

   I have performed ANOVA  anayalsis of ALFF among PD depression group (Group 1), PD non-depression group (Group 2) and control group (Group 3). I chose AlpahSim for multiple correction under the voxel p value of 0.001 and cluster p value of 0.05. After ANOVA, I performed post-hoc under the mask file of F.nii other than the Z value abtained by DPABI, for the reason that I wanted to add disease duration as covariates when compared Group 1 and Group 2. Am I right of the method? 

   Using the above method, I found that when comparing to Group 3, both Group 1 and Group 2 had different ALFF value in SMA. However, SMA was not found when performing ANOVA analysis. It seems strange and unreasonable to me. I wonder if I was wrong somehow, or was it just normal?

   Hope for your reply.

  Best!

   Mia