micapipe from the Multimodal imaging and connectome analysis lab (http://mica-mni.github.io) at the Montreal Neurological Institute. Read The Docs documentation below
I hope this email finds you well. I am writing to inquire about the dimension of the matrix data in the output of the Structural Connectome module of the micapipe toolkit (v 0.1.4 & 0.1.5).
Specifically, I have noticed that in the micapipe//dwi/connectomes directory, there were files named _space-dwi_atlas-desc-iFOD2--SIFT2-connectome.txt, which I expected to contain a matrix of size 100100 for the Schaefer2018_100Parcels cortical template. However, upon inspection, the actual matrix size is 150150, with the first 49 rows/columns being empty. I understand that this is intended to store the connectivity data for the subcortical, corresponding to the full_connectome.txt. Starting from the 50th row/column or nodes, actual connectivity data is present, resulting in an effective matrix size of 101101. Similar observations have been made with other cortical templates, such as the glasser360 template, which has a matrix size of 361361. Interestingly, the aparc-a2009s template, which should yield a matrix size of 150150 (or 198198, with the first 48 rows/columns being empty), appears to have a consistent matrix size with its cortical labels' number.
I have a few questions regarding these observations:
Questions 1:
For templates like the 100 template, the one extra node that exceed the expected size of the cortical template (e.g., 101101 instead of 100100), does it represent the "unknown" label for cortex? If so, does this additional node correspond to the first or the last node in the 101*101 matrix?
Questions 3:
Additionally, for the aparc-a2009s template, which seems to have only 48 nodes in the matrix. In my understanding, the number of subcortical nodes obtained using FreeSurfer preprocessed cortical segmentation should be fixed for the same subject.
Answers:
This maybe be due to an issue with the labels file, we noticed this issue before and fixed it on our latest release v0.2.0.
Questions 4:
I am aware that the aparc-a2009s template also contains "unknown" cortex labels in the template parcellation, in addition to the 150 labels already named. However, I am puzzled as to why the actual data for this template aligns with the expected matrix size of 150*150, while other templates yield an additional node in the matrix. This inconsistency has left me confused.
Thank Dr. Raul Rodriguezcruces for the answers.
The original content of the email:
I hope this email finds you well. I am writing to inquire about the dimension of the matrix data in the output of the Structural Connectome module of the micapipe toolkit (v 0.1.4 & 0.1.5).
Specifically, I have noticed that in the micapipe//dwi/connectomes directory, there were files named _space-dwi_atlas-desc-iFOD2--SIFT2-connectome.txt, which I expected to contain a matrix of size 100100 for the Schaefer2018_100Parcels cortical template. However, upon inspection, the actual matrix size is 150150, with the first 49 rows/columns being empty. I understand that this is intended to store the connectivity data for the subcortical, corresponding to the full_connectome.txt. Starting from the 50th row/column or nodes, actual connectivity data is present, resulting in an effective matrix size of 101101. Similar observations have been made with other cortical templates, such as the glasser360 template, which has a matrix size of 361361. Interestingly, the aparc-a2009s template, which should yield a matrix size of 150150 (or 198198, with the first 48 rows/columns being empty), appears to have a consistent matrix size with its cortical labels' number.
I have a few questions regarding these observations:
Questions 1: For templates like the 100 template, the one extra node that exceed the expected size of the cortical template (e.g., 101101 instead of 100100), does it represent the "unknown" label for cortex? If so, does this additional node correspond to the first or the last node in the 101*101 matrix?
Answers: For more cases this extra label represents the middle wall. you can find the LUT (look up tables) of all the parcellations in here: https://github.com/MICA-MNI/micapipe/tree/master/parcellations/lut.
Questions 2: Are the first 49 nodes in the matrix data subcortical? If yes, where can I find the corresponding label information for these nodes?
Answers: Here: https://github.com/MICA-MNI/micapipe/blob/master/parcellations/lut/lut_subcortical-cerebellum_mics.csv
Questions 3: Additionally, for the aparc-a2009s template, which seems to have only 48 nodes in the matrix. In my understanding, the number of subcortical nodes obtained using FreeSurfer preprocessed cortical segmentation should be fixed for the same subject.
Answers: This maybe be due to an issue with the labels file, we noticed this issue before and fixed it on our latest release v0.2.0.
Questions 4: I am aware that the aparc-a2009s template also contains "unknown" cortex labels in the template parcellation, in addition to the 150 labels already named. However, I am puzzled as to why the actual data for this template aligns with the expected matrix size of 150*150, while other templates yield an additional node in the matrix. This inconsistency has left me confused.
Answers: This is the aparc-a2009 LUT: https://github.com/MICA-MNI/micapipe/blob/master/parcellations/lut/lut_aparc-a2009s_mics.csv