Unfortunately I can't give you a simple answer, because the file format for field maps changes with _scanner_ software version, and the apparent intensity scaling depends on the software _you_ are using.
Philips sometimes writes field maps as [real TE1, imag TE1, real TE2, imag TE2], but for ADNI3 they are coming out as [mag TE1, phase TE1, mag TE2, phase TE2] as you said. Technically the intensity scaling on the phase images is set in the DICOM headers, but in practice I convert to .nii as soon as possible, and different versions of dcm2niix produce different scalings, either -pi to pi, or (-pi, pi) * 2e8! Either way, though the images are wrapped phase images, so whatever the scaling is, the actual minimum and maximum values are really (-pi, pi), so you can solve for the scaling.
You will need deltaTE = TE2 - TE1, and you can get the Echo Times from the DICOM headers of the 1st and 2nd set.
I hope this helps.