Encoding the local connectivity patterns of fMRI for cognitive task and state classification

2019-08-01
Ertugrul, Itir Onal
Ozay, Mete
Yarman Vural, Fatoş Tunay
In this work, we propose a novel framework to encode the local connectivity patterns of brain, using Fisher vectors (FV), vector of locally aggregated descriptors (VLAD) and bag-of-words (BoW) methods. We first obtain local descriptors, called mesh arc descriptors (MADs) from fMRI data, by forming local meshes around anatomical regions, and estimating their relationship within a neighborhood. Then, we extract a dictionary of relationships, called brain connectivity dictionary by fitting a generative Gaussian mixture model (GMM) to a set of MADs, and selecting codewords at the mean of each component of the mixture. Codewords represent connectivity patterns among anatomical regions. We also encode MADs by VLAD and BoW methods using k-Means clustering. We classify cognitive tasks using the Human Connectome Project (HCP) task fMRI dataset and cognitive states using the Emotional Memory Retrieval (EMR). We train support vector machines (SVMs) using the encoded MADs. Results demonstrate that, FV encoding of MADs can be successfully employed for classification of cognitive tasks, and outperform VLAD and BoW representations. Moreover, we identify the significant Gaussians in mixture models by computing energy of their corresponding FV parts, and analyze their effect on classification accuracy. Finally, we suggest a new method to visualize the codewords of the learned brain connectivity dictionary.
BRAIN IMAGING AND BEHAVIOR

Suggestions

Rapid multi-orientation quantitative susceptibility mapping
Bilgic, Berkin; Xie, Luke; Dibb, Russell; Langkammer, Christian; Mutluay, Aysegul; Ye, Huihui; Polimeni, Jonathan R.; Augustinack, Jean; Liu, Chunlei; Wald, Lawrence L.; Setsompop, Kawin (Elsevier BV, 2016-01-15)
Three-dimensional gradient echo (GRE) is the main workhorse sequence used for susceptibility weighted imaging (SWI), quantitative susceptibility mapping (QSM), and susceptibility tensor imaging (STI). Achieving optimal phase signal-to-noise ratio requires late echo times, thus necessitating a long repetition time (TR). Combined with the large encoding burden of whole-brain coverage with high resolution, this leads to increased scan time. Further, the dipole kernel relating the tissue phase to the underlying...
Local image structures and optic flow estimation
Kalkan, Sinan; Worgotter, F.; Lappe, M.; Kruger, N. (Informa UK Limited, 2005-12-01)
Different kinds of local image structures (such as homogeneous, edge-like and junction-like patches) can be distinguished by the intrinsic dimensionality of the local signals. Intrinsic dimensionality makes use of variance from a point and a line in spectral representation of the signal in order to classify it as homogeneous, edge-like or junction-like. The concept of intrinsic dimensionality has been mostly exercised using discrete formulations; however, recent work (Felsberg & Kriger 2003; Kruger & Felsbe...
First-order and second-order statistical analysis of 3d and 2d image structure
Kalkan, Sinan; Kruger, N. (Informa UK Limited, 2007-06-01)
In the first part of this article, we analyze the relation between local image structures (i.e., homogeneous, edge-like, corner-like or texture-like structures) and the underlying local 3D structure (represented in terms of continuous surfaces and different kinds of 3D discontinuities) using range data with real-world color images. We find that homogeneous image structures correspond to continuous surfaces, and discontinuities are mainly formed by edge-like or corner-like structures, which we discuss regard...
Identification of nonlinear features in cortical and subcortical signals of Parkinson's Disease patients via a novel efficient measure
Özkurt, Tolga Esat; Zrinzo, Ludvic; Limousin, Patricia; Foltynie, Tom; Oswal, Ashwini; Litvak, Vladimir (Elsevier BV, 2020-12-01)
This study offers a novel and efficient measure based on a higher order version of autocorrelative signal memory that can identify nonlinearities in a single time series. The suggested method was applied to simultaneously recorded subthalamic nucleus (STN) local field potentials (LFP) and magnetoencephalography (MEG) from fourteen Parkinson's Disease (PD) patients who underwent surgery for deep brain stimulation. Recordings were obtained during rest for both OFF and ON dopaminergic medication states. We ana...
Generation of cyclic/toroidal chaos by Hopfield neural networks
Akhmet, Marat (Elsevier BV, 2014-12-05)
We discuss the appearance of cyclic and toroidal chaos in Hopfield neural networks. The theoretical results may strongly relate to investigations of brain activities performed by neurobiologists. As new phenomena, extension of chaos by entrainment of several limit cycles as well as the attraction of cyclic chaos by an equilibrium are discussed. Appropriate simulations that support the theoretical results are depicted. Stabilization of tori in a chaotic attractor is realized not only for neural networks, but...
Citation Formats
I. O. Ertugrul, M. Ozay, and F. T. Yarman Vural, “Encoding the local connectivity patterns of fMRI for cognitive task and state classification,” BRAIN IMAGING AND BEHAVIOR, pp. 893–904, 2019, Accessed: 00, 2020. [Online]. Available: https://hdl.handle.net/11511/39435.