If you assume the LNT (theory A) the cumulative effects of the dose at Fukushima on the surrounding population might be a 0.1% increase in cancer deaths over what would be expected. Given that there are 100,000 people in the vicinity, that might be 100 extra deaths (pulling numbers out of my backside here, but they are plausible to within an order of magnitude).
The trouble is that a sample size of 100,000 isn't enough to reliably demonstrate a 0.1% increase in cancer rates, in the same way that tossing a coin 100,000 times isn't enough to reliably demonstrate that a coin comes up heads 50.1% of the time rather than 50.
There's no way in the world we'll ever get this kind of data from human studies absent global nuclear war, in which case we'll have more important things to worry about. The only plausible way you might useful data would be a very, very large scale animal study, probably costing many millions of dollars.