The problem is not so much the strong magnetic field (either a huge permanent magnet or a super conducting magnet), because a high temperature super conducting magnet might be feasible in the (near?) future.
It's not the hf coils for reading the returning signals.
It's homogeneity! An mri device works by reading the energy at a specific frequency level which reaches the hf reading coils. The frequency describes the location of origin because gradient coils are used to superimpose a slight offset onto the basic field which varies by location. So that each point in the target volume has a different magnetic field strength. That way the lamor (or lamour?) Frequency changes and the signal received by the reading coils has a different frequency for each location of origin. That's the frequency-space coding that allows you to reconstruct an image.
The less precise the magnetic field, the lesser the image quality. There are open bore mris and small mri for small probes, but shooting a magnetic field around and getting any kind of signal that is more than nonsense is not feasible in near future.
So let's go further in time. If you can measure very low energy and effectively calculate all the noise you would get, the presence of a human in the magnetic field of the earth could be picked up as slightly different noise by a reading coil.
Combine that with a high temperature super conducting magnet which generates a magnetic field at a target (handwave, or maybe someday by adding cancellation) or a visual measurement of the distance and the expected magnetic field there and a lot of calculation power and you could actually get a signal from another human.
As to blur, the quality of the image depends on measuring time and homogeneity. If however you could calculate exactly how the field would look like at your target and what deviations you'd get and compensate motion with the camera data etc. You could get some good data.
However the calculation necessary is way out of our reach today and if you had an ai capable of that, there would probably be easier ways to calculate the needed dosage.
For example, a simple rgb camera can already measure the heart rate of someone by filming their skin (the green channel is best for just reading out the waves from the bloodflow). Combine that with an infrared camera, some statistical data and a weight and fitness estimate and you can already get a lot of information about the physical condition.
TL;DR: no, a lot would have to happen first, essentially leaving you in a future with very high tech and a lot of different possibilities we can only imagine today.