Investigating Community Detection Algorithms and their Capacity as Markers of Brain Diseases
| Authors | |
|---|---|
| Year of publication | 2017 |
| Type | Article in Proceedings |
| Conference | International Symposium on Grids and Clouds (ISGC) 2017. Academia Sinica, Taipei, Taiwan: Proceedings of Science |
| MU Faculty or unit | |
| Citation | |
| Doi | http://dx.doi.org/10.22323/1.293.0018 |
| Keywords | Classification (of information); Optimization; Population dynamics; Random variables |
| Description | In this paper, we present a workflow for evaluating resting-state brain functional connectivity with different community detection algorithms and their strengths to discriminate between health and Parkinson’s disease (PD) and mild cognitive impairment preceding Alzheimer’s disease (ADMCI). We further analyze the complexity of particular pipeline steps aiming to provide guidelines for both execution on computing infrastructure and further optimization efforts. On a dataset of 50 controls and 70 patients we measured an increased modularity coefficient with 81.8% accuracy of classifying PD versus controls and 76.2% accuracy of classifying ADMCI versus controls. Significantly higher modularity coefficient values were measured when the random matrix theory decomposition was adapted for network construction. These results were observed on networks of 82 nodes based on AAL atlas and 317 nodes based on multimodal parcellation atlas. |
| Related projects: |