External Email - Use Caution
Dear Freesurfer’s experts,
I have read the paper from Marcoux, et al. « An Automated Pipeline for the Analysis of PET Data on the Cortical Surface ». Frontiers in Neuroinformatics 12 (10 décembre 2018). https://doi.org/10.3389/fninf.2018.00094 https://doi.org/10.3389/fninf.2018.00094. In this paper, they propose a robust PET signal projection based on weighting multiple PET projections at different fractions of the cortical thickness. They used « mris_expand" instead of "mri_vol2surf --projfrac" based on the following comments:
"We mainly rely on the mris_expand function of FreeSurfer rather than on vol2surf. The vol2surf function allows the user to project volume data onto the cortical surface by sampling along the normal of the white surface, by steps corresponding to fractions of the cortical thickness. We observed that the main issue with this approach is its lack of accuracy and robustness, as there is no constraint to ensure that the last sample will be on the corresponding vertex of the pial surface. Another solution proposed by FreeSurfer is to use the normal fromthemid surface, starting on the side of the white surface and going toward the pial surface, sampling by steps proportional to the thickness. Here again we observed that, because of different folding patterns, sampling points could be located outside the cortical ribbon. Our approach, using deformable surfaces obtained with mris_expand, ensures that the seven surfaces that are used to project the signal follow the folding patterns and stay within the cortical ribbon."
Is "mris_expand" more accurate and robust than "mri_vol2surf --projfrac" according to you ?
Best, Matthieu