binarybottle
commented
6 years ago When Mindboggle's write_shape_stats creates tables of summary statistics (median, mad, mean, sd, etc.) of the vertex measures (curvature, depths, etc.), the stats_per_label function normalizes by area. This ensures that the shape of flat regions don't drive the summary statistics of the shape of a region due to the nonuniform distribution of vertices on the surface meshes. This normalization by surface area should be the cause of the differences between the two numbers you are generating.
Hi,
Using defaults we calculate for subject 'Arno' ("mindboggle_input_example") mindboggle output.
Our objective is to compare output-table values from mindboggle with values we calculated outside mindboggle but using the mindboggle input.
Outside mindboggle, for subject Arno, we calculate mean geodesic depth of the left fundus of the superior frontal sulcus by taking the average of the values of "shapes/left_cortical_surface/geodesic_depth.vtk" along the indexed points of the superior frontal sulcus (ID = 1, found in the "features/left_cortical_surface/fundus_per_sulcus.vtk"). Depth is 12.2772
The output-table geodesic depth from mindboggle ("mindboggle_input_example/freesurfer/subjects/mindboggle_output/arno/tables/left_cortical_surface/fundus_shapes.csv) gives for "geode sic depth: mean": 10.444352911051064
The question: What are we (outside mindboggle) doing differently compared to 'inside' mindboggle? One thing that comes to mind is a correction of some sort in mindboggle (akin to what freesurfer does) and including the corrected values in the output-tables. What are the inputs and what is the formula of this correction?
Thanks! -joost