TABLE 1.
Connectome Preprocessing
Graph Measure | Edge Weight | Preprocessing | ||||
---|---|---|---|---|---|---|
Number of Streamlines | Mean Length | Threshold | Normalize | Binarize | Remove NaN/∞ | |
Modularity | X | |||||
Global efficiency | X | X | ||||
Assortativity | X | |||||
Density | X | X | ||||
Density with adaptive thresholding | X | X | X | |||
Edge count | X | X | ||||
Characteristic path length | X | |||||
Small-worldness | X | X | ||||
Randomness | X | X | ||||
Average participation coefficient | X | |||||
Average local efficiency | X | X | ||||
Average local efficiency with adaptive thresholding | X | X | X | |||
Average node strength | X | |||||
Average normalized node strength | X | X | ||||
Average node degree | X | |||||
Average normalized node degree | X | X | ||||
Average clustering coefficient | X | X | X | |||
Average clustering coefficient with adaptive thresholding | X | X | X | X | ||
Average betweenness centrality | X |
Detailed here are the edge weight and preprocessing steps associated with each graph measure. Each connectome is preprocessed with BCT. Density, local efficiency, node strength, and clustering coefficient have alternate versions in which additional preprocessing is applied. We offer alternative definitions that converge to an unbiased estimate, where the original definitions do not.