Looking at Picture2:

The EPI is upside down. This indicates that for whatever reason, it's z-axis encoded as I-S, 82.2mm S to 177.2mm S, is backwards. The SPGR is displayed as AFNI should show them (of course, the L-R distinction is hard to tell from here, but looks OK since the R hemisphere goes anterior a little more than the left, and the L hemisphere goes a little farther posterior than the right, as is common among us Homo Sapiens types).

You can either try to figure out why the EPI is upside down and shifted off to lie entirely in the Superior half of the world, or you can fix it after its creation via to3d. You can fix it with the program '3drefit', using with the option '-orient RAS'. The directions of the coordinates may be OK then, but it is likely the location in space will not overlap the anatomy well. Depending on how badly they don't line up, you might have to take further steps, such as using the @Align_Centers script for crude registration and/or the align_epi_anat.py script for fine registration of EPI to SPGR.

Did you create this EPI dataset using Rick Reynold's DIMON program? If that will work for you, it's the best approach. However, we don't have a lot of experience with Philips DICOM files, so perhaps there are some problems we aren't familiar with.

Finally, I see that you are using a Mac. I thought Taiwan was a PC-only land. Did you smuggle this computer into the island?