Maybe late, but I thought I'd contribute since I have experience with MRI engineering.
It depends on what you mean by "non-magnetic". If you're just concerned with the components becoming missiles in the MRI's field, then you need to avoid large solid bodies of ferrous material. But small amounts (relative to the mass of the whole device) can be fine (so long as they're securely fixed in place). Many connectors are commonly available with solid brass bodies, and are safe to use in MRI applications. I've used SMA, SMB, SMC, and MCX at fields up to 3T. I've also used BNC, but only specific parts which I knew were safe (I think we used 11_BNC-50-2-13/133_NE from Huber $uhner).
However, most of those connectors I used did have some amount of ferrous material, most often nickel underplating (extremely common underneath gold plated contacts). The overall mass of the nickel is very small relative to the rest of the part, so it won't turn into a missile in a strong field. However that will still disqualify it from being sold as a "non-magnetic" part.
Some other things to be aware of:
1. Always check your components and assemblies for excessive magnetic attraction before introducing them to the MRI.
2. Many connectors have stainless steel parts. A simple google search will often suggest that SS is non-magnetic. Never, ever take this at face value, even if they claim it's an Austenitic alloy or something. See #1 above.
3. Many coaxial cables use copper clad steel (CSC) as the inner conductor. This is definitely not good for MRI. Usually this is determined by the cable's standard, for example all RG-58 should have solid copper, while all RG-316 should have CSC. Best to check the datasheet for the specific cable you buy.
4. Even if the nickel plating isn't enough to turn the parts into missiles, they may distort the B0 field enough to degrade image quality, depending on how close they are to the imaging FOV. Usually this is pretty obvious, but it depends on what sort of sequence/application you're working with...