FiberNET: An Ensemble Deep Learning Framework for Clustering White Matter Fibers

Gupta, Vikash; Thomopoulos, Sophia I.; Rashid, Faisal; Thompson, Paul M.

doi:10.1007/978-3-319-66182-7_63

Cited by 28 publications

(14 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Creating fiber bundle templates requires manual segmentation because of the huge number of spurious fiber tracts present in the tractogram, thus making the process expensive. However, in the future, we intend to use FiberNet [12], a deep learning tool for automatic extraction of fiber bundles. Using this tool, it will become feasible to construct a population template using a larger sample of individuals.…”

Section: Discussionmentioning

confidence: 99%

White matter differences in Parkinson’s disease mapped using tractometry

Corbin

Gupta

Villalon‐Reina

et al. 2017

Preprint

Self Cite

View full text Add to dashboard Cite

Neurodegenerative disorders are characterized by a progressive loss of brain function. Improved precision in mapping the altered brain pathways can provide a deep understanding of the trajectory of decline. We propose a tractometry workflow for conducting group statistical analyses of pointwise microstructural measures along white matter fasciculi to identify patterns of abnormalities associated with disease. We combined state-of-the-art tools including fiber registration, tract simplification and fiber matching for accurate point-wise statistical analyses across populations. We test the utility of this method by identifying group differences between Parkinson's disease (PD) patients and healthy controls. We find statistically significant group differences in diffusion MRI derived measures along the anterior thalamic radiations (ATR), corticospinal tract (CST) and regions of the corpus callosum (CC). These pathways are essential for motor control systems within cortico-cortical and cortico-subcortical brain networks. Moreover, the reported pathological changes were not widespread but rather localized along several tracts. Point-wise tract analyses may therefore offer an advantage in anatomical specificity over traditional methods that assess mean microstructural measures across large regions of interest.

show abstract

Section: Discussionmentioning

confidence: 99%

White matter differences in Parkinson’s disease mapped using tractometry

Corbin

Gupta

Villalon‐Reina

et al. 2017

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Inspired by the success of convolutional neural networks in the computer vision community [4], recently CNNs have begun to be tested for clustering white matter fibers in the brain. In this paper, we extend work we presented in [5], to better handle false positive fibers. A whole brain tractography algorithm produces as much as four times the false-positive fibers (pathways that are incorrectly tracked and do not correspond to true anatomy) when compared to true fibers [6].…”

Section: Introductionmentioning

confidence: 86%

“…These parameters vary across different datasets and scanners. To resolve this, we use a volumetric parameterization technique, proposed in [5]. This parameterization may also be thought of as a normalization process for the fiber tracts.…”

Section: Volumetric Parameterizationmentioning

confidence: 99%

“…In this paper, we extended the workflow described in a previous version of FiberNET [5] for classifying WM fibers. These changes help us counter the presence of spurious fibers, which have been said to comprise a majority of fiber tracts computed using typical tractography algorithms [6].…”

Section: Fibernet 20mentioning

confidence: 99%

“…To this end, false positive fibers cause serious problems for fiber clustering methods, and for any downstream analyses based on tractography. In our prior work [5], we were not able to compensate for spurious fibers, but we presented a proof of concept that CNN based neural network architectures might be used for fiber clustering tasks. In this paper, we build a more practical and robust approach to reduce the effects of false positive streamlines.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Fibernet 2.0: An Automatic Neural Network Based Tool for Clustering White Matter Fibers in the Brain

Gupta

Thomopoulos

Corbin

et al. 2017

Preprint

Self Cite

View full text Add to dashboard Cite

The brain's white matter fiber tracts are impaired in a range of common and devastating conditions, from Alzheimer's disease to brain trauma, and in developmental disorders such as autism and neurogenetic syndromes. Many studies now examine the connectivity and microstructure of the brain's neural pathways, spurring the development of algorithms to extract and measure tracts and fiber bundles. Clustering white matter (WM) fibers, from whole-brain tractography, into anatomically meaningful bundles is still a challenging problem. Existing tract segmentation methods use atlases or regions of interest (ROI) or unsupervised spectral clustering. Even so, atlas-based segmentation does not always partition the brain into a set of recognizable fiber bundles. Deep learning techniques can be applied to automatically segment and cluster white matter fibers. Here we propose a robust approach using convolutional neural networks (CNNs) to learn shape features of the fiber bundles, which we then exploit to cluster WM fibers into bundles. In a range of tests across diverse fiber bundles, we illustrate the accuracy of our method, and its ability to suppress false positive fibers.

show abstract