Understanding the MBM Output

The MBM (mouse-build-model) pipeline outputs three csv files containing pointers to processed files necessary for statistical analysis.

---

If 11.1_coil15_distcorr.mnc is the name of the input volume file, [VOLUME] denotes 11.1_coil15_distcorr.

transforms.csv contains 10 columns

• lsq12_nlin_xfm <- [VOLUME]_N_I_lsq6_lsq12_and_nlin.xfm The 3x4 matrix contains the 3x3 lsq12 (does not include lsq6) scale, shear, and rotation matrix; augmented with the 3x1 translation vector. The grid transform contains the nlin transform.

• overall_xfm <- [VOLUME]__concat_lsq6_N_I_lsq6_lsq12_and_nlin.xfm The 3x4 matrix contains the product of the 3x3 lsq6 and lsq12 scale, shear, and rotation matrix; augmented with the sum of their 3x1 translation vectors. The grid transform contains the nlin transform.

• rigid_xfm <- [VOLUME]_lsq6.xfm The 3x4 matrix contains the 3x3 lsq6 rotation matrix, augmented with the 3x1 translation vector.

• xfm_to_common <- concat_of_[VOLUME]_N_I_lsq6_lsq12_and_nlin_and_nlin-3_to_second_level-nlin-3_lsq12_ANTS_nlin.xfm This transform takes the file in the lsq6 space to the common space.

• overall_xfm_to_common <- concat_of_[VOLUME]_lsq6_and_[VOLUME]_N_I_lsq6_lsq12_and_nlin_and_nlin-3_to_second_level-nlin-3_lsq12_ANTS_nlin.xfm This transform is the concatenation of the lsq6 transform and the xfm_to_common` transform.

• native_file <- [VOLUME].mnc This is the native input minc file.

• lsq6_file <- [VOLUME]_N_I_lsq6.mnc This is the minc file after having undergone non-uniformity correction, intensity normalization (image processing steps); and lsq6 transformation (rotation, and displacement).

• lsq6_mask_file <- [VOLUME]_N_I_lsq6_max_mask.mnc This is the mask for lsq6_file created by applying the transformation required to register the input atlas to the native file, on the input atlas`s mask.

• nlin_file <- [VOLUME]_N_I_lsq6_lsq12_and_nlin-resampled.mnc This is the minc file after having undergone non-uniformity correction, intensity normalization, lsq6, lsq12 and nlin registration.

• common_space_file <- concat_of_[VOLUME]_N_I_lsq6_lsq12_and_nlin_and_nlin-3_to_second_level-nlin-3_lsq12_ANTS_nlin-resampled.mnc This is the minc file re-sampled into the common space.

determinants.csv contains 7 columns and is not tidy

• fwhm <- [FWHM] This is the full-width-half-maximum value of the isotropic 3D Gaussian blurring kernel in mm. For each distinct value of fwhm, the rows are duplicated.

• inv_xfm <- [VOLUME]_N_I_lsq6_lsq12_and_nlin.xfm This is the same as lsq12_nlin_xfm of transforms.csv.

• log_full_det <- [VOLUME]_N_I_lsq6_lsq12_and_nlin_inverted_displ_log_det_abs_fwhm[FWHM].mnc This minc file contains the logarithm of the determinant of the lsq12 and nlin transformation in nlin space.

• log_nlin_det <- [VOLUME]_N_I_lsq6_lsq12_and_nlin__concat_inverted_linear_part_displ_log_det_rel_fwhm[FWHM].mnc This minc file contains the logarithm of the determinant of the nlin transformation in nlin space.

• xfm <- [VOLUME]_N_I_lsq6_lsq12_and_nlin_inverted.xfm This is the inverse of inv_xfm or lsq12_nlin_xfm of transforms.csv—the transform from nlin space to lsq6 space.

• log_nlin_det_common <- [VOLUME]_N_I_lsq6_lsq12_and_nlin__concat_inverted_linear_part_displ_log_det_rel_fwhm[FWHM]_res_to_second_level-nlin-3_common.mnc This minc file is log_nlin_det resampled into the common space.

• log_full_det_common <- [VOLUME]_N_I_lsq6_lsq12_and_nlin_inverted_displ_log_det_abs_fwhm[FWHM]_res_to_second_level-nlin-3_common.mnc This minc file is log_full_det resampled into the common space.

analysis.csv

analysis.csv is a convenient catchall database that relates the input features, native filenames and output filenames necessary for analysis. It is the join of input.csv, transforms.csv, and determinants.csv with the xfm and inv_xfm columns from determinants.csv dropped. It has extra columns pointing to the label files resulting from maget and the nlin-resampled masks.