What a fantastic question. It depends on the type of imaging used.
Structural MRI is what we call it when you use the MRI scanner to take a detailed 3D picture of the brain as illustrated on my profile and in my blog. As you can see, this gets an accurate picture of the whole head, with no gaps. This is great for telling what size and shape bits of the brain are, but it doesn’t tell us what they’re doing,
fMRI, which is the main method for looking at which parts of the brain are active in different tasks has some problems imaging brain tissue which is very near to air-filled parts of the head (called sinuses – you’d be surprised how much air there is inside your head, I was at least). The scanner takes pictures of the brain in very quick slices (1 every ~0.1s) and the effect of this problem depends a little on which direction the slices go in. The main problems are found in the anterior temporal lobes. If you look at a picture of the brain from side on, each side looks a little like a boxing glove (the front being the fist). The anterior temporal lobes are like the tips of the thumbs.
Parts of the brain are also somewhat distorted in fMRI (because the huge magnetic field we use is not quite “even”). In my experience this seems to affect the front and underside of the frontal lobes (the fist in the boxing glove analogy).
You still get measurements from these parts of the brain, but they are weaker and less reliable.
Another technique we sometimes use is MEG. This uses an entirely different scanner which measures the tiny magnetic fields generated when brain cells work together. The sensors (which I think are something like the detectors Marieke uses in her work, but I am not sure) are arranged around the head in a very large and expensive hat (filled with liquid helium). They can detect signals generated by the outside of the brain nearest the sensors, but have a harder time picking up signals from deeper inside.