Thalamus parcellation using multi-modal feature classification and thalamic nuclei priors

Jeffrey Glaister, Aaron Carass, Joshua V. Stough, Peter Calabresi, Jerry Ladd Prince

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

Abstract

Segmentation of the thalamus and thalamic nuclei is useful to quantify volumetric changes from neurodegenerative diseases. Most thalamus segmentation algorithms only use T1-weighted magnetic resonance images and current thalamic parcellation methods require manual interaction. Smaller nuclei, such as the lateral and medial geniculates, are challenging to locate due to their small size. We propose an automated segmentation algorithm using a set of features derived from diffusion tensor image (DTI) and thalamic nuclei location priors. After extracting features, a hierarchical random forest classifier is trained to locate the thalamus. A second random forest classifies thalamus voxels as belonging to one of six thalamic nuclei classes. The proposed algorithm was tested using a leave-one-out cross validation scheme and compared with state-of-the-art algorithms. The proposed algorithm has a higher Dice score compared to other methods for the whole thalamus and several nuclei.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2016: Image Processing
PublisherSPIE
Volume9784
ISBN (Electronic)9781510600195
DOIs
StatePublished - 2016
EventMedical Imaging 2016: Image Processing - San Diego, United States
Duration: Mar 1 2016Mar 3 2016

Other

OtherMedical Imaging 2016: Image Processing
CountryUnited States
CitySan Diego
Period3/1/163/3/16

Fingerprint

thalamus
Thalamic Nuclei
Thalamus
nuclei
Neurodegenerative diseases
Magnetic resonance
classifiers
Neurodegenerative Diseases
Tensors
magnetic resonance
Classifiers
Magnetic Resonance Spectroscopy
tensors

Keywords

  • Diffiusion MRI
  • Machine learning
  • Magnetic resonance imaging
  • Segmentation
  • Thalamus parcellation

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics
  • Electronic, Optical and Magnetic Materials
  • Biomaterials
  • Radiology Nuclear Medicine and imaging

Cite this

Glaister, J., Carass, A., Stough, J. V., Calabresi, P., & Prince, J. L. (2016). Thalamus parcellation using multi-modal feature classification and thalamic nuclei priors. In Medical Imaging 2016: Image Processing (Vol. 9784). [97843J] SPIE. https://doi.org/10.1117/12.2216987

Thalamus parcellation using multi-modal feature classification and thalamic nuclei priors. / Glaister, Jeffrey; Carass, Aaron; Stough, Joshua V.; Calabresi, Peter; Prince, Jerry Ladd.

Medical Imaging 2016: Image Processing. Vol. 9784 SPIE, 2016. 97843J.

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

Glaister, J, Carass, A, Stough, JV, Calabresi, P & Prince, JL 2016, Thalamus parcellation using multi-modal feature classification and thalamic nuclei priors. in Medical Imaging 2016: Image Processing. vol. 9784, 97843J, SPIE, Medical Imaging 2016: Image Processing, San Diego, United States, 3/1/16. https://doi.org/10.1117/12.2216987
Glaister J, Carass A, Stough JV, Calabresi P, Prince JL. Thalamus parcellation using multi-modal feature classification and thalamic nuclei priors. In Medical Imaging 2016: Image Processing. Vol. 9784. SPIE. 2016. 97843J https://doi.org/10.1117/12.2216987
Glaister, Jeffrey ; Carass, Aaron ; Stough, Joshua V. ; Calabresi, Peter ; Prince, Jerry Ladd. / Thalamus parcellation using multi-modal feature classification and thalamic nuclei priors. Medical Imaging 2016: Image Processing. Vol. 9784 SPIE, 2016.
@inproceedings{ff72cc441cc14aa89ca76011a56a577e,
title = "Thalamus parcellation using multi-modal feature classification and thalamic nuclei priors",
abstract = "Segmentation of the thalamus and thalamic nuclei is useful to quantify volumetric changes from neurodegenerative diseases. Most thalamus segmentation algorithms only use T1-weighted magnetic resonance images and current thalamic parcellation methods require manual interaction. Smaller nuclei, such as the lateral and medial geniculates, are challenging to locate due to their small size. We propose an automated segmentation algorithm using a set of features derived from diffusion tensor image (DTI) and thalamic nuclei location priors. After extracting features, a hierarchical random forest classifier is trained to locate the thalamus. A second random forest classifies thalamus voxels as belonging to one of six thalamic nuclei classes. The proposed algorithm was tested using a leave-one-out cross validation scheme and compared with state-of-the-art algorithms. The proposed algorithm has a higher Dice score compared to other methods for the whole thalamus and several nuclei.",
keywords = "Diffiusion MRI, Machine learning, Magnetic resonance imaging, Segmentation, Thalamus parcellation",
author = "Jeffrey Glaister and Aaron Carass and Stough, {Joshua V.} and Peter Calabresi and Prince, {Jerry Ladd}",
year = "2016",
doi = "10.1117/12.2216987",
language = "English (US)",
volume = "9784",
booktitle = "Medical Imaging 2016: Image Processing",
publisher = "SPIE",

}

TY - GEN

T1 - Thalamus parcellation using multi-modal feature classification and thalamic nuclei priors

AU - Glaister, Jeffrey

AU - Carass, Aaron

AU - Stough, Joshua V.

AU - Calabresi, Peter

AU - Prince, Jerry Ladd

PY - 2016

Y1 - 2016

N2 - Segmentation of the thalamus and thalamic nuclei is useful to quantify volumetric changes from neurodegenerative diseases. Most thalamus segmentation algorithms only use T1-weighted magnetic resonance images and current thalamic parcellation methods require manual interaction. Smaller nuclei, such as the lateral and medial geniculates, are challenging to locate due to their small size. We propose an automated segmentation algorithm using a set of features derived from diffusion tensor image (DTI) and thalamic nuclei location priors. After extracting features, a hierarchical random forest classifier is trained to locate the thalamus. A second random forest classifies thalamus voxels as belonging to one of six thalamic nuclei classes. The proposed algorithm was tested using a leave-one-out cross validation scheme and compared with state-of-the-art algorithms. The proposed algorithm has a higher Dice score compared to other methods for the whole thalamus and several nuclei.

AB - Segmentation of the thalamus and thalamic nuclei is useful to quantify volumetric changes from neurodegenerative diseases. Most thalamus segmentation algorithms only use T1-weighted magnetic resonance images and current thalamic parcellation methods require manual interaction. Smaller nuclei, such as the lateral and medial geniculates, are challenging to locate due to their small size. We propose an automated segmentation algorithm using a set of features derived from diffusion tensor image (DTI) and thalamic nuclei location priors. After extracting features, a hierarchical random forest classifier is trained to locate the thalamus. A second random forest classifies thalamus voxels as belonging to one of six thalamic nuclei classes. The proposed algorithm was tested using a leave-one-out cross validation scheme and compared with state-of-the-art algorithms. The proposed algorithm has a higher Dice score compared to other methods for the whole thalamus and several nuclei.

KW - Diffiusion MRI

KW - Machine learning

KW - Magnetic resonance imaging

KW - Segmentation

KW - Thalamus parcellation

UR - http://www.scopus.com/inward/record.url?scp=84981738875&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84981738875&partnerID=8YFLogxK

U2 - 10.1117/12.2216987

DO - 10.1117/12.2216987

M3 - Conference contribution

VL - 9784

BT - Medical Imaging 2016: Image Processing

PB - SPIE

ER -