If I may chime on on this. It is a very interesting area to work on but to my knowledge
there are no published solutions. When we think of BOLD sensitivity there are a host of
factors - physiological and instrumental - that play parts. Because, for example, the
vascular reactivity may differ by region, some of these contributors are deeply buried.
Nevertheless, some thoughts might help. It is well accepted that a paramagnetic field
pertuber (e.g., deoxyHgb) will decrease in R2* (1/T2*) that is approximately linear with
concentration. It is not too difficult to show that this will propagate to the observed signal
intensity in such a way that the fractional signal change (% signal change) is very close to
linear over a broad concentration range and over a broad range of tissue T2* values.
Therefore, measured as % signal change, the BOLD sensitivity to [deoxyHgb] is close to
uniform. However, the _statistical_ sensitivity is another matter.
In MRI, the instrumentation noise is more or less constant for all locations. Thus, the signal
to noise ratio, and therefore the detection sensitivity is roughly proportional to the absolute
image intensity. Finally, you should be aware that in most cases, the noise per voxel is (and
ideally should be) dominated by physiologically-based signal fluctuations. One crude test for
this would be to calculate the variance (or std dev) by region. If the noise is dominated by
physiology, the stdev should be approximately proportional to the signal intensity.
I am working on the manuscript, but I hope this helps in the meantime.