A novel graph neural network to localize eloquent cortex in brain tumor patients from resting-state FMRI connectivity

Naresh Nandakumar, Komal Manzoor, Jay J. Pillai, Sachin K. Gujar, Haris I. Sair, Archana Venkataraman

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We develop a novel method to localize the language and motor areas of the eloquent cortex in brain tumor patients based on resting-state fMRI (rs-fMRI) connectivity. Our method leverages the representation power of convolutional neural networks through specialized filters that act topologically on the rs-fMRI connectivity data. This Graph Neural Network (GNN) classifies each parcel in the brain into eloquent cortex, tumor, or background gray matter, thus accommodating varying tumor characteristics across patients. Our loss function also reflects the large class-imbalance present in our data. We evaluate our GNN on rs-fMRI data from 60 brain tumor patients with different tumor sizes and locations. We use motor and language task fMRI for validation. Our model achieves better localization than linear SVM, random forest, and a multilayer perceptron architecture. Our GNN is able to correctly identify bilateral language areas in the brain even when trained on patients whose language network is lateralized to the left hemisphere.

Original languageEnglish (US)
Title of host publicationConnectomics in NeuroImaging - 3rd International Workshop, CNI 2019, Held in Conjunction with MICCAI 2019, Proceedings
EditorsMarkus D. Schirmer, Ai Wern Chung, Archana Venkataraman, Islem Rekik, Minjeong Kim
PublisherSpringer
Pages10-20
Number of pages11
ISBN (Print)9783030323905
DOIs
StatePublished - Jan 1 2019
Event3rd International Workshop on Connectomics in NeuroImaging, CNI 2019 held in conjunction with the 22nd International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2019 - Shenzhen, China
Duration: Oct 13 2019Oct 13 2019

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11848 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference3rd International Workshop on Connectomics in NeuroImaging, CNI 2019 held in conjunction with the 22nd International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2019
CountryChina
CityShenzhen
Period10/13/1910/13/19

Fingerprint

Brain Tumor
Functional Magnetic Resonance Imaging
Cortex
Tumors
Brain
Connectivity
Neural Networks
Neural networks
Tumor
Graph in graph theory
Random Forest
Hemisphere
Loss Function
Perceptron
Leverage
Multilayer
Multilayer neural networks
Classify
Filter
Language

Keywords

  • Graph Neural Network
  • Language localization
  • Rs-fMRI

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Nandakumar, N., Manzoor, K., Pillai, J. J., Gujar, S. K., Sair, H. I., & Venkataraman, A. (2019). A novel graph neural network to localize eloquent cortex in brain tumor patients from resting-state FMRI connectivity. In M. D. Schirmer, A. W. Chung, A. Venkataraman, I. Rekik, & M. Kim (Eds.), Connectomics in NeuroImaging - 3rd International Workshop, CNI 2019, Held in Conjunction with MICCAI 2019, Proceedings (pp. 10-20). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11848 LNCS). Springer. https://doi.org/10.1007/978-3-030-32391-2_2

A novel graph neural network to localize eloquent cortex in brain tumor patients from resting-state FMRI connectivity. / Nandakumar, Naresh; Manzoor, Komal; Pillai, Jay J.; Gujar, Sachin K.; Sair, Haris I.; Venkataraman, Archana.

Connectomics in NeuroImaging - 3rd International Workshop, CNI 2019, Held in Conjunction with MICCAI 2019, Proceedings. ed. / Markus D. Schirmer; Ai Wern Chung; Archana Venkataraman; Islem Rekik; Minjeong Kim. Springer, 2019. p. 10-20 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11848 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Nandakumar, N, Manzoor, K, Pillai, JJ, Gujar, SK, Sair, HI & Venkataraman, A 2019, A novel graph neural network to localize eloquent cortex in brain tumor patients from resting-state FMRI connectivity. in MD Schirmer, AW Chung, A Venkataraman, I Rekik & M Kim (eds), Connectomics in NeuroImaging - 3rd International Workshop, CNI 2019, Held in Conjunction with MICCAI 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11848 LNCS, Springer, pp. 10-20, 3rd International Workshop on Connectomics in NeuroImaging, CNI 2019 held in conjunction with the 22nd International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2019, Shenzhen, China, 10/13/19. https://doi.org/10.1007/978-3-030-32391-2_2
Nandakumar N, Manzoor K, Pillai JJ, Gujar SK, Sair HI, Venkataraman A. A novel graph neural network to localize eloquent cortex in brain tumor patients from resting-state FMRI connectivity. In Schirmer MD, Chung AW, Venkataraman A, Rekik I, Kim M, editors, Connectomics in NeuroImaging - 3rd International Workshop, CNI 2019, Held in Conjunction with MICCAI 2019, Proceedings. Springer. 2019. p. 10-20. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-32391-2_2
Nandakumar, Naresh ; Manzoor, Komal ; Pillai, Jay J. ; Gujar, Sachin K. ; Sair, Haris I. ; Venkataraman, Archana. / A novel graph neural network to localize eloquent cortex in brain tumor patients from resting-state FMRI connectivity. Connectomics in NeuroImaging - 3rd International Workshop, CNI 2019, Held in Conjunction with MICCAI 2019, Proceedings. editor / Markus D. Schirmer ; Ai Wern Chung ; Archana Venkataraman ; Islem Rekik ; Minjeong Kim. Springer, 2019. pp. 10-20 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{a613e601cddd4f6c80d7da95b72b8f70,
title = "A novel graph neural network to localize eloquent cortex in brain tumor patients from resting-state FMRI connectivity",
abstract = "We develop a novel method to localize the language and motor areas of the eloquent cortex in brain tumor patients based on resting-state fMRI (rs-fMRI) connectivity. Our method leverages the representation power of convolutional neural networks through specialized filters that act topologically on the rs-fMRI connectivity data. This Graph Neural Network (GNN) classifies each parcel in the brain into eloquent cortex, tumor, or background gray matter, thus accommodating varying tumor characteristics across patients. Our loss function also reflects the large class-imbalance present in our data. We evaluate our GNN on rs-fMRI data from 60 brain tumor patients with different tumor sizes and locations. We use motor and language task fMRI for validation. Our model achieves better localization than linear SVM, random forest, and a multilayer perceptron architecture. Our GNN is able to correctly identify bilateral language areas in the brain even when trained on patients whose language network is lateralized to the left hemisphere.",
keywords = "Graph Neural Network, Language localization, Rs-fMRI",
author = "Naresh Nandakumar and Komal Manzoor and Pillai, {Jay J.} and Gujar, {Sachin K.} and Sair, {Haris I.} and Archana Venkataraman",
year = "2019",
month = "1",
day = "1",
doi = "10.1007/978-3-030-32391-2_2",
language = "English (US)",
isbn = "9783030323905",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer",
pages = "10--20",
editor = "Schirmer, {Markus D.} and Chung, {Ai Wern} and Archana Venkataraman and Islem Rekik and Minjeong Kim",
booktitle = "Connectomics in NeuroImaging - 3rd International Workshop, CNI 2019, Held in Conjunction with MICCAI 2019, Proceedings",

}

TY - GEN

T1 - A novel graph neural network to localize eloquent cortex in brain tumor patients from resting-state FMRI connectivity

AU - Nandakumar, Naresh

AU - Manzoor, Komal

AU - Pillai, Jay J.

AU - Gujar, Sachin K.

AU - Sair, Haris I.

AU - Venkataraman, Archana

PY - 2019/1/1

Y1 - 2019/1/1

N2 - We develop a novel method to localize the language and motor areas of the eloquent cortex in brain tumor patients based on resting-state fMRI (rs-fMRI) connectivity. Our method leverages the representation power of convolutional neural networks through specialized filters that act topologically on the rs-fMRI connectivity data. This Graph Neural Network (GNN) classifies each parcel in the brain into eloquent cortex, tumor, or background gray matter, thus accommodating varying tumor characteristics across patients. Our loss function also reflects the large class-imbalance present in our data. We evaluate our GNN on rs-fMRI data from 60 brain tumor patients with different tumor sizes and locations. We use motor and language task fMRI for validation. Our model achieves better localization than linear SVM, random forest, and a multilayer perceptron architecture. Our GNN is able to correctly identify bilateral language areas in the brain even when trained on patients whose language network is lateralized to the left hemisphere.

AB - We develop a novel method to localize the language and motor areas of the eloquent cortex in brain tumor patients based on resting-state fMRI (rs-fMRI) connectivity. Our method leverages the representation power of convolutional neural networks through specialized filters that act topologically on the rs-fMRI connectivity data. This Graph Neural Network (GNN) classifies each parcel in the brain into eloquent cortex, tumor, or background gray matter, thus accommodating varying tumor characteristics across patients. Our loss function also reflects the large class-imbalance present in our data. We evaluate our GNN on rs-fMRI data from 60 brain tumor patients with different tumor sizes and locations. We use motor and language task fMRI for validation. Our model achieves better localization than linear SVM, random forest, and a multilayer perceptron architecture. Our GNN is able to correctly identify bilateral language areas in the brain even when trained on patients whose language network is lateralized to the left hemisphere.

KW - Graph Neural Network

KW - Language localization

KW - Rs-fMRI

UR - http://www.scopus.com/inward/record.url?scp=85075681632&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85075681632&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-32391-2_2

DO - 10.1007/978-3-030-32391-2_2

M3 - Conference contribution

AN - SCOPUS:85075681632

SN - 9783030323905

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 10

EP - 20

BT - Connectomics in NeuroImaging - 3rd International Workshop, CNI 2019, Held in Conjunction with MICCAI 2019, Proceedings

A2 - Schirmer, Markus D.

A2 - Chung, Ai Wern

A2 - Venkataraman, Archana

A2 - Rekik, Islem

A2 - Kim, Minjeong

PB - Springer

ER -