Brain abnormality segmentation based on l 1-norm minimization

Ke Zeng, Guray Erus, Manoj Tanwar, Christos Davatzikos

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

Abstract

We present a method that uses sparse representations to model the inter-individual variability of healthy anatomy from a limited number of normal medical images. Abnormalities in MR images are then defined as deviations from the normal variation. More precisely, we model an abnormal (pathological) signal y as the superposition of a normal part ~y that can be sparsely represented under an example-based dictionary, and an abnormal part r. Motivated by a dense error correction scheme recently proposed for sparse signal recovery, we use l1- norm minimization to separate ~y and r. We extend the existing framework, which was mainly used on robust face recognition in a discriminative setting, to address challenges of brain image analysis, particularly the high dimensionality and low sample size problem. The dictionary is constructed from local image patches extracted from training images aligned using smooth transformations, together with minor perturbations of those patches. A multi-scale sliding-window scheme is applied to capture anatomical variations ranging from fine and localized to coarser and more global. The statistical significance of the abnormality term r is obtained by comparison to its empirical distribution through cross-validation, and is used to assign an abnormality score to each voxel. In our validation experiments the method is applied for segmenting abnormalities on 2-D slices of FLAIR images, and we obtain segmentation results consistent with the expert-defined masks.

Original languageEnglish (US)
Title of host publicationProgress in Biomedical Optics and Imaging - Proceedings of SPIE
PublisherSPIE
Volume9034
ISBN (Print)9780819498274
DOIs
StatePublished - 2014
Externally publishedYes
EventMedical Imaging 2014: Image Processing - San Diego, CA, United States
Duration: Feb 16 2014Feb 18 2014

Other

OtherMedical Imaging 2014: Image Processing
CountryUnited States
CitySan Diego, CA
Period2/16/142/18/14

Fingerprint

abnormalities
Glossaries
norms
brain
Brain
optimization
Error correction
Masks
Face recognition
dictionaries
Sample Size
Image analysis
Anatomy
Recovery
anatomy
image analysis
Experiments
sliding
education
masks

Keywords

  • Abnormality segmentation
  • Brain MRI
  • Convex optimization
  • Sparse representation

ASJC Scopus subject areas

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

Cite this

Zeng, K., Erus, G., Tanwar, M., & Davatzikos, C. (2014). Brain abnormality segmentation based on l 1-norm minimization. In Progress in Biomedical Optics and Imaging - Proceedings of SPIE (Vol. 9034). [903409] SPIE. https://doi.org/10.1117/12.2043146

Brain abnormality segmentation based on l 1-norm minimization. / Zeng, Ke; Erus, Guray; Tanwar, Manoj; Davatzikos, Christos.

Progress in Biomedical Optics and Imaging - Proceedings of SPIE. Vol. 9034 SPIE, 2014. 903409.

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

Zeng, K, Erus, G, Tanwar, M & Davatzikos, C 2014, Brain abnormality segmentation based on l 1-norm minimization. in Progress in Biomedical Optics and Imaging - Proceedings of SPIE. vol. 9034, 903409, SPIE, Medical Imaging 2014: Image Processing, San Diego, CA, United States, 2/16/14. https://doi.org/10.1117/12.2043146
Zeng K, Erus G, Tanwar M, Davatzikos C. Brain abnormality segmentation based on l 1-norm minimization. In Progress in Biomedical Optics and Imaging - Proceedings of SPIE. Vol. 9034. SPIE. 2014. 903409 https://doi.org/10.1117/12.2043146
Zeng, Ke ; Erus, Guray ; Tanwar, Manoj ; Davatzikos, Christos. / Brain abnormality segmentation based on l 1-norm minimization. Progress in Biomedical Optics and Imaging - Proceedings of SPIE. Vol. 9034 SPIE, 2014.
@inproceedings{0e688ba0d1a1481da656e3ef2977d629,
title = "Brain abnormality segmentation based on l 1-norm minimization",
abstract = "We present a method that uses sparse representations to model the inter-individual variability of healthy anatomy from a limited number of normal medical images. Abnormalities in MR images are then defined as deviations from the normal variation. More precisely, we model an abnormal (pathological) signal y as the superposition of a normal part ~y that can be sparsely represented under an example-based dictionary, and an abnormal part r. Motivated by a dense error correction scheme recently proposed for sparse signal recovery, we use l1- norm minimization to separate ~y and r. We extend the existing framework, which was mainly used on robust face recognition in a discriminative setting, to address challenges of brain image analysis, particularly the high dimensionality and low sample size problem. The dictionary is constructed from local image patches extracted from training images aligned using smooth transformations, together with minor perturbations of those patches. A multi-scale sliding-window scheme is applied to capture anatomical variations ranging from fine and localized to coarser and more global. The statistical significance of the abnormality term r is obtained by comparison to its empirical distribution through cross-validation, and is used to assign an abnormality score to each voxel. In our validation experiments the method is applied for segmenting abnormalities on 2-D slices of FLAIR images, and we obtain segmentation results consistent with the expert-defined masks.",
keywords = "Abnormality segmentation, Brain MRI, Convex optimization, Sparse representation",
author = "Ke Zeng and Guray Erus and Manoj Tanwar and Christos Davatzikos",
year = "2014",
doi = "10.1117/12.2043146",
language = "English (US)",
isbn = "9780819498274",
volume = "9034",
booktitle = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",
publisher = "SPIE",

}

TY - GEN

T1 - Brain abnormality segmentation based on l 1-norm minimization

AU - Zeng, Ke

AU - Erus, Guray

AU - Tanwar, Manoj

AU - Davatzikos, Christos

PY - 2014

Y1 - 2014

N2 - We present a method that uses sparse representations to model the inter-individual variability of healthy anatomy from a limited number of normal medical images. Abnormalities in MR images are then defined as deviations from the normal variation. More precisely, we model an abnormal (pathological) signal y as the superposition of a normal part ~y that can be sparsely represented under an example-based dictionary, and an abnormal part r. Motivated by a dense error correction scheme recently proposed for sparse signal recovery, we use l1- norm minimization to separate ~y and r. We extend the existing framework, which was mainly used on robust face recognition in a discriminative setting, to address challenges of brain image analysis, particularly the high dimensionality and low sample size problem. The dictionary is constructed from local image patches extracted from training images aligned using smooth transformations, together with minor perturbations of those patches. A multi-scale sliding-window scheme is applied to capture anatomical variations ranging from fine and localized to coarser and more global. The statistical significance of the abnormality term r is obtained by comparison to its empirical distribution through cross-validation, and is used to assign an abnormality score to each voxel. In our validation experiments the method is applied for segmenting abnormalities on 2-D slices of FLAIR images, and we obtain segmentation results consistent with the expert-defined masks.

AB - We present a method that uses sparse representations to model the inter-individual variability of healthy anatomy from a limited number of normal medical images. Abnormalities in MR images are then defined as deviations from the normal variation. More precisely, we model an abnormal (pathological) signal y as the superposition of a normal part ~y that can be sparsely represented under an example-based dictionary, and an abnormal part r. Motivated by a dense error correction scheme recently proposed for sparse signal recovery, we use l1- norm minimization to separate ~y and r. We extend the existing framework, which was mainly used on robust face recognition in a discriminative setting, to address challenges of brain image analysis, particularly the high dimensionality and low sample size problem. The dictionary is constructed from local image patches extracted from training images aligned using smooth transformations, together with minor perturbations of those patches. A multi-scale sliding-window scheme is applied to capture anatomical variations ranging from fine and localized to coarser and more global. The statistical significance of the abnormality term r is obtained by comparison to its empirical distribution through cross-validation, and is used to assign an abnormality score to each voxel. In our validation experiments the method is applied for segmenting abnormalities on 2-D slices of FLAIR images, and we obtain segmentation results consistent with the expert-defined masks.

KW - Abnormality segmentation

KW - Brain MRI

KW - Convex optimization

KW - Sparse representation

UR - http://www.scopus.com/inward/record.url?scp=84902094474&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84902094474&partnerID=8YFLogxK

U2 - 10.1117/12.2043146

DO - 10.1117/12.2043146

M3 - Conference contribution

AN - SCOPUS:84902094474

SN - 9780819498274

VL - 9034

BT - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

PB - SPIE

ER -