Spherical ROI on surface mesh

Software Support
, ,
1

Hello,
I have been struggling overlaying a self defined ROI on a brain surface with a statistical map. For my analysis I need to mask left DLPFC area (representing correlations), which I have done
image

Next, I would like to add a spherical ROI on this image at a specific coordinate. It sort of works on 2D image
image

But I get weird results plotting it on a mesh
image

The code I have used:

masked_data = apply_mask(seedmap_img, dlpfc_image)
    
masked_img = unmask(masked_data, dlpfc_image)

texture = surface.vol_to_surf(masked_img, struct_surf)

roi_map = roi_img.get_fdata()

from nilearn.surface import vol_to_surf
roi_surf = vol_to_surf(roi_img,surf_mesh=struct_surf)

figure=plot_surf_stat_map(
    struct_surf, texture,
    hemi='left',
    view='lateral',
    colorbar=True,
    alpha=1,
    threshold=None,          
    cmap='cold_hot',             
    bg_on_data=False,            
    )

plot_surf_roi(
    surf_mesh = struct_surf,
    roi_map = roi_surf,
    bg_map = texture,
    threshold=None,
    figure=figure,
    bg_on_data=True,
)
plt.show()

I have tried changing or deleting plot_surf_stat_map and plot_surf_roi arguments, but this is the best image I got so far. Does anyone know how to get the desired overlay?

Best,
Vytautas

2

I guess that this comes from the fact that fsaverage does not follow well the brain outline in frontal regions. So when you project heavily masked data to fsaverage, the result is not great. TO me it is weird to perform ROI definition and on the surface, and then the display on a mesh. You should project the data, and then specify the ROI and apply the masking on the surface directly. This would be more consistent.
HTH,
Bertrand