Dear Asmin, The discrepancies arise from differences in anatomical protocols. You can look at the definitions in the corresponding papers and keep the one that adapts better to your needs. The results should be fairly highly correlated. Kind regards, Eugenio

-- Juan Eugenio Iglesias http://www.jeiglesias.com

From: freesurfer-bounces@nmr.mgh.harvard.edu on behalf of Asmin Alam asmin.alam23@gmail.com Reply-To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Date: Monday, March 13, 2023 at 13:02 To: "freesurfer@nmr.mgh.harvard.edu" freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] Hypothalamus segmentation using hypothalamic subunits vs ScLimbic

External Email - Use Caution Dear FreeSurfer Experts,

I have a question regarding how the hypothalamus volumes are calculated using the hypothalamic subunits function, as opposed to the ScLimbic deep learning tool which is only currently available in the dev version of Freesurfer.

My project consists of measuring the volume of the hypothalamus, so I thought of running both algorithms and comparing results to ensure they were consistent. However, they are quite different and I was wondering if there are certain flags/options I should be adding to ensure that the results will relate to each other and why that is the case. What is the key difference in how the hypothalamus is calculated by both these methods? How do I ensure my results are robust/trustworthy? What does noMB mean in the ScLimbic results?

I have tried running the ScLimbic function using the original T1 and the --conform flag to make sure that the voxels were isotropic, and then retried by using the nu.mgz as input (as that's what the hypothalamus subunits function uses). Here are the results (also attached):

HYPOTHALAMIC SUBUNITS subject left anterior-inferior left anterior-superior left posterior left tubular inferior left tubular superior right anterior-inferior right anterior-superior right posterior right tubular inferior right tubular superior whole left whole right test 19.88 25.155 136.715 145.785 131.879 20.078 29.209 134.882 161.009 133.293 459.414 478.471

SC LIMBIC (T1-W and --conform) # Subcortical Limbic Volumetric Stats # Created by mri_sclimbic_seg # NRows 12 # NTableCols 5 # ColHeaders Index SegId NVoxels Volume_mm3 StructName 1 26 469 471.6122 Left-Nucleus-Accumbens 2 58 460 459.8144 Right-Nucleus-Accumbens 3 819 588 571.3680 Left-HypoThal-noMB 4 820 631 586.4990 Right-HypoThal-noMB 5 821 667 707.3562 Left-Fornix 6 822 631 700.7991 Right-Fornix 7 843 58 60.5358 Left-MammillaryBody 8 844 57 64.1979 Right-MammillaryBody 9 865 414 396.2163 Left-Basal-Forebrain 10 866 448 450.1535 Right-Basal-Forebrain 11 869 165 170.6947 Left-SeptalNuc 12 870 151 157.4303 Right-SeptalNuc

SC LIMBIC (nu.mgz input) # Subcortical Limbic Volumetric Stats # Created by mri_sclimbic_seg # NRows 12 # NTableCols 5 # ColHeaders Index SegId NVoxels Volume_mm3 StructName 1 26 513 519.7247 Left-Nucleus-Accumbens 2 58 526 537.3985 Right-Nucleus-Accumbens 3 819 575 570.7261 Left-HypoThal-noMB 4 820 613 604.3893 Right-HypoThal-noMB 5 821 657 675.3337 Left-Fornix 6 822 663 691.8721 Right-Fornix 7 843 51 53.4853 Left-MammillaryBody 8 844 58 61.4891 Right-MammillaryBody 9 865 473 483.7591 Left-Basal-Forebrain 10 866 487 496.4204 Right-Basal-Forebrain 11 869 145 153.3073 Left-SeptalNuc 12 870 138 142.8245 Right-SeptalNuc

I'm also not sure exactly what the flag "--crop" means for the hypothalamic subunits function. I left it to the default value of 184 but how will changing this affect my final result?

Below are also my codes for all three functions: mri_segment_hypothalamic_subunits --s ${subjid} --sd ${sd} --crop 184 mri_sclimbic_seg --i ${sd}/${subjid}/mri//T1.mgz --o ${sd}/${subjid}/mri/sclimbic.mgz --write_volumes --conform mri_sclimbic_seg --i ${sd}/${subjid}/mri//nu.mgz --o ${sd}/${subjid}/mri/sclimbic.mgz --write_volumes

Any help is appreciated. Thank you, Asmin

UCL Queen Square Institute of Neurology