Human Claustrum connections: robust in-vivo detection by DWI-based tractography in two large samples

Annotations for: Human Claustrum connections: robust in-vivo detection by DWI-based tractography in two large samples

✅ Key Points

deep-learning-based claustrum 35 segmentation tool with DWI-based tractography in two large adult cohorts: 81 healthy young adults 36 from the human connectome project and 81 further healthy young participants from the Bavarian 37 longitudinal study. [@wendtHumanClaustrumConnections, p. 3]

consistent and replicable cortical 42 and subcortical ipsi- and contralateral claustrum connections. [@wendtHumanClaustrumConnections, p. 3]

his study deals with the consistency and replicability of diffusion-weighted imaging (DWI)51 based reconstruction of claustrum connections via probabilistic tractography in the human brain. [@wendtHumanClaustrumConnections, p. 4]

Annotation entry (Type: 2, Color: #ffd400) [@wendtHumanClaustrumConnections, p. 4]

Milardi et al. provided further evidence of large fiber tracts between 104 the claustrum and most cortical areas, as well as of the existence of contralateral projections and 105 interclaustral communications bundles (Milardi et al., 2015; Arrigo et al., 2017). [@wendtHumanClaustrumConnections, p. 5]

Misspelled linear [@wendtHumanClaustrumConnections, p. 12]

linar [@wendtHumanClaustrumConnections, p. 12]

Background

Quite a nice sentence, here [@wendtHumanClaustrumConnections, p. 4]

It has been suggested that 76 the basic physiology of claustrum-cortex circuits, namely selective inhibitory suppression of the cortex, 77 makes the claustrum suited for distributed but coordinated cortex control, which is in service of the 78 aforementioned distinct functions (Jackson et al., 2020; Smith et al., 2020). [@wendtHumanClaustrumConnections, p. 4]

Importantly for human research, the sum of 93 findings from different animals, show marked differences across species (Jackson et al., 2020), 94 suggesting that those findings may not be transferable to humans without reservations. [@wendtHumanClaustrumConnections, p. 5]

For voxel size, HCP had a higher resolution at 1.25 mm3 isotropic 436 when compared to 2 mm3 isotropic for BLS. [@wendtHumanClaustrumConnections, p. 13]

Hypothesis

81 further non-related healthy individuals were taken from the control arm of the Bavarian 153 Longitudinal Study (Wolke and Meyer, 1999), which investigates neonatal at-risk children and healthy 154 term-born controls recruited at birth in southern Bavaria, Germany, between January 1985 and March 155 1986. [@wendtHumanClaustrumConnections, p. 6]

Methods

it remains unclear if this indirect method is really able to accurately and – most importantly for the 112 current study – robustly (i.e., consistently and replicably) reconstruct the extent of the claustrum’s 113 connections, amongst other open questions. [@wendtHumanClaustrumConnections, p. 5]

inexactitudes due to a small number of subjects 114 as well as in high-yield graph-theoretical analysis using standardized and not anatomical parcellations 115 are likely (Welton et al., 2015). [@wendtHumanClaustrumConnections, p. 5]

sample size was addressed by analysing a large human sample, imaged and processed using 126 state-of-the-art DWI and quality control protocols (Glasser et al., 2013; Sotiropoulos et al., 2013). [@wendtHumanClaustrumConnections, p. 5]

Findings were reproduced in a second large sample with similar mean age and gender distribution, 128 imaged using imaging protocols more similar to those used in clinical routines (Wolke and Meyer, [@wendtHumanClaustrumConnections, p. 5]

1999; Meng et al., 2016; Menegaux et al., 2021). [@wendtHumanClaustrumConnections, p. 6]

probabilistic tractography approach (Behrens et al., 2007) [@wendtHumanClaustrumConnections, p. 6]

individual claustrum 133 masks obtained with a deep learning-based automated claustrum segmentation tool developed in-house 134 (Li et al., 2021; Neubauer et al., 2022). [@wendtHumanClaustrumConnections, p. 6]

Results

preprocessed using the initial stages of the PreQual pipeline (Cai et al., 2021), 225 which included MP-PCA denoising, inter-scan normalization, as well as generation of a synthetic 226 undistorted b0 using the SynB0-DisCo deep-learning framework (Schilling et al., 2019; Schilling et al., 227 2020), an additional step necessary in this case as no reverse phase-encoded images were available to 228 use for distortion correction. [@wendtHumanClaustrumConnections, p. 8]

After extracting the brain from the surrounding tissue using FSL BET 231 (Smith, 2002), FSL EDDY (Andersson and Sotiropoulos, 2016) was used for eddy current-induced 232 distortion and inter- and intra-volume movement correction, with the options to replace outliers 233 (Andersson et al., 2016) and to perform slice-wise signal dropout imputation (Andersson et al., 2017) [@wendtHumanClaustrumConnections, p. 8]

group-wise and single-subject quantitative quality metrics were 255 automatically derived from the data using FSL QUAD and SQUAD (Bastiani et al. 2019). [@wendtHumanClaustrumConnections, p. 9]

why?!! [@wendtHumanClaustrumConnections, p. 12]

SPSS statistics package version 28 (IBM) [@wendtHumanClaustrumConnections, p. 12]

Conclusions

here i would remove the word supplements [@wendtHumanClaustrumConnections, p. 15]

(see Table S4 in supplements for 517 detailed streamline data) [@wendtHumanClaustrumConnections, p. 15]

Implications

double brackets) [@wendtHumanClaustrumConnections, p. 8]

(Marcus et al., 2013) (van Essen et al., 2013); [@wendtHumanClaustrumConnections, p. 8]

This difference needs to be taken 392 into account when examining differing results, as white matter connectivity is significantly influenced 393 by age and changes during development (Lebel et al., 2012). [@wendtHumanClaustrumConnections, p. 12]

Access Required