I suspect the very low contrast between the fossil and the surrounding rock would mean that either we need a very sensitive sensor, very long exposures or likely both.
Ground penetrating radar can be used to visualize fossils but the conductivity of the material greatly impacts resolution (it's poor no matter what) and reach. Low-conductivity materials can be imaged up to tens of meters, high-conductivity materials you're lucky to get one meter.
There are other methods for imaging that can penetrate further but I don't think they have the resolution to be useful (think: "there's an oil deposit down there" not "there's a body").
Routinely scanning random chunks of stone would be prohibitively expensive, of course.
A number of quarries already automate the slicing and "inspect" surfaces for defects via computer vision.
The trick is pattern recognition to catch things of interest so that if required the raw slices can be retrieved before they're shipped.
It's destructive tomography .. with an option to digitally reconstruct a solid, or even to physically rejoin slices.
How about neutrino imaging? (I thought I was joking, but... https://www.nature.com/articles/s41567-018-0319-1)