Not really an "answer", but it will depend upon what you want to do with the data/inferences you wish to make. Do you expect each scan to give you "complimentary" windows on the brain? This make require some special consideration and knowledge of your scanner setup(s).

Easiest thing might be to analyse data together at the statistical analysis phase, including scanner type as nuisance regressor. But again, this depends upon need.

It is not clear from your question whether this is to be an intra-subject combination or an inter-subject combination. The change in field from 1.5 to 3 T does give greater clarity and contrast in antomical brain images. This comes with a difference in artifacts at each field. You will not be able to directly compare the data from the two scans but you may be able to use measures of different structures. The lower SNR or lower resolution at 1.5T will show up as a change in how good your surface image is rendered in your analysis software and will give you a larger error compared to the 3T images.

Dear John,

Based on the application, the utility of 1.5T and 3T varies. As morris, said 3T provides better clarity in understanding the anatomical images for Ex: medial temporal cortex.

Could you tell me what your exactly looking for?

Thanks

Generally, no - not for quantitative analysis. You can't directly combine any two different MRI sequences - unless they are specifically designed to complement each other. In the simplest case: if structural clinical images are constructed from several samples of free induction decay (FID) curve (in reality, you work with a superposition of FID curves), so your tissue contrast totally depends on the combination of sequence pulses, sample timing and chosen contrast capture mechanism (T1w, T2w, etc). Bore magnet strength also affects the shape of the FID curve, so even if you deploy identical sequences on 1.5 and 3T magnets, you can't expect to get comparable contrasts.

Having said that, if one could predict the FID curves (their superposition, that is) for the tissue of interest, one could combine them for multiple sequences, with the result valid only for this particular tissue type.

can't be used for image fusion to my knowledge

Yes, If you model it as a multisite study with a variable coding for the site.

Thanks all for your helpful input.

Not due to software privacy and image can be fused with same type of software.