I love the way the maxima plugin now automatically numbers regions, but I'm a little confused about how the weighted average option for dealing with neighboring extrema works. These are the 17 maxima I obtained:
RAI mm coordinates:
( -17.00 33.00 30.00) : val = 2.730000
( -32.00 57.00 14.00) : val = 4.440000
( 27.00 47.00 33.00) : val = 2.950000
( -21.00 -43.00 14.00) : val = 2.850000
( 52.00 23.00 -3.00) : val = 2.030000
( -45.00 27.00 35.00) : val = 6.180000
( -22.00 -24.00 11.00) : val = 15.980000
( 43.00 -27.00 19.00) : val = 4.060000
( 11.00 31.00 24.00) : val = 1.740000
( 23.00 63.00 7.00) : val = 13.770000
( -42.00 48.00 0.00) : val = 12.220000
( 25.00 16.00 49.00) : val = 0.820000
( -18.00 8.00 49.00) : val = 11.870000
( -53.00 35.00 49.00) : val = 11.860000
( 24.00 65.00 42.00) : val = 11.780000
( -19.00 14.00 28.00) : val = 11.630000
( -13.00 8.00 -3.00) : val = 10.960000
number of extrema = 17
I was initially confused by values that were well below my threshold, but since the weighted average option may assign the local maximum to a voxel that isn't as intensely activated as the maxima surrounding it, I think I understand this now. My question is about how these are ranked in cases where extrema are clustered. Is it based on the most intense extreme within the cluster, or do all the extrema within the cluster impact the ranking? Would it be possible to list the values that determine the ranking rather than the value of the actual center of the cluster?
While we're at it, what are the advantages and disadvantages of "nearest neighbor" vs. "weighted average"?