@inproceedings{edd8732c625242a486037b8a72da74a5,
title = "Product space decompositions for continuous representations of brain connectivity",
abstract = "We develop a method for the decomposition of structural brain connectivity estimates into locally coherent components, leveraging a non-parametric Bayesian hierarchical mixture model with tangent Gaussian components. This model provides a mechanism to share information across subjects while still including explicit mixture distributions of connections for each subject. It further uses mixture components defined directly on the surface of the brain, eschewing the usual graph-theoretic framework of structural connectivity in favor of a continuous model that avoids a priori assumptions of parcellation configuration. The results of two experiments on a test-retest dataset are presented, to validate the method. We also provide an example analysis of the components.",
keywords = "Brain connectivity, Non-parametric bayes, Unsupervised learning",
author = "Daniel Moyer and Gutman, {Boris A.} and Neda Jahanshad and Thompson, {Paul M.}",
note = "Publisher Copyright: {\textcopyright} 2017, Springer International Publishing AG.; 8th International Workshop on Machine Learning in Medical Imaging, MLMI 2017 held in conjunction with the 20th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2017 ; Conference date: 10-09-2017 Through 10-09-2017",
year = "2017",
doi = "10.1007/978-3-319-67389-9_41",
language = "English",
isbn = "9783319673882",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "353--361",
editor = "Yinghuan Shi and Heung-Il Suk and Kenji Suzuki and Qian Wang",
booktitle = "Machine Learning in Medical Imaging - 8th International Workshop, MLMI 2017, Held in Conjunction with MICCAI 2017, Proceedings",
address = "Germany",
}