Gast ::
| Hauptseite > Publikationsdatenbank > Painting the Brain by Numbers: Introducing an open-source approach to automated regional connectivity-based parcellation |
| Hauptseite > Publikationsdatenbank > Painting the Brain by Numbers: Introducing an open-source approach to automated regional connectivity-based parcellation |
| Dissertation / PhD Thesis | FZJ-2022-05889 |
2022
Please use a persistent id in citations: http://hdl.handle.net/2128/33147
Abstract: Regional connectivity-based parcellation (rCBP) is a widely used procedure for investigating the structural and functional differentiation within a region-of-interest (ROI) based on its long-range connectivity. No standardized software or guidelines currently exist for applying rCBP, making the method only accessible to those who develop their own tools. A historical background to rCBP has been provided in chapter 1, which continues with the aim of this work: introducing CBPtools, an open-source software package implementing rCBP. The chapter concludes by detailing various methods and concepts associated with the rCBP procedure.CBPtools is a Python (version 3.5+) package that allows users to run an extensively evaluated rCBP analysis workflow on a given ROI. It currently supports two modalities: resting-state functional connectivity and structural connectivity based on diffusion-weighted imaging, along with support for custom connectivity matrices. Analysis parameters are customizable, and the workflow can be scaled to many subjects using a parallel processing environment. Parcellation results with corresponding validity metrics are provided as textual and graphical output. Thus, CBPtools provides a simple plug-and-play yet customizable way to conduct rCBP analyses. Chapter two discusses architectural choices, scope, and software dependencies, followed by a thorough description of all implemented features as well as a step-by-step guide through the processing pipeline.In chapter three we demonstrate the utility of CBPtools using a voluminous data set on an average compute-cluster infrastructure by performing rCBP on three ROIs prominently featured in parcellation literature. A side-project on the investigation of potential issues regarding outliers in the data set is added as chapter four.In closing we discuss our findings, provide recommendations, and suggest future extensions to the CBPtools software in chapters five and six. CBPtools is capable of reproducing parcellations found in existing literature. It offers flexibility in terms of customization while remaining easy to use. By providing an open-source software we aim to promote reproducible and comparable rCBP analyses and, importantly, make the rCBP procedure readily available.
|
The record appears in these collections: |
PDF
Fulltext by OpenAccess repository