Semi-automated segmentation of pre-operative low grade gliomas in magnetic resonance imaging

Zeynettin Akkus, Jiri Sedlar, Lucie Coufalova, Panagiotis Korfiatis, Timothy Kline, Joshua D. Warner, Jay Agrawal, Bradley J Erickson

Research output: Contribution to journalArticle

10 Citations (Scopus)

Abstract

Background: Segmentation of pre-operative low-grade gliomas (LGGs) from magnetic resonance imaging is a crucial step for studying imaging biomarkers. However, segmentation of LGGs is particularly challenging because they rarely enhance after gadolinium administration. Like other gliomas, they have irregular tumor shape, heterogeneous composition, ill-defined tumor boundaries, and limited number of image types. To overcome these challenges we propose a semi-automated segmentation method that relies only on T2-weighted (T2W) and optionally post-contrast T1-weighted (T1W) images. Methods: First, the user draws a region-of-interest (ROI) that completely encloses the tumor and some normal tissue. Second, a normal brain atlas and post-contrast T1W images are registered to T2W images. Third, the posterior probability of each pixel/voxel belonging to normal and abnormal tissues is calculated based on information derived from the atlas and ROI. Finally, geodesic active contours use the probability map of the tumor to shrink the ROI until optimal tumor boundaries are found. This method was validated against the true segmentation (TS) of 30 LGG patients for both 2D (1 slice) and 3D. The TS was obtained from manual segmentations of three experts using the Simultaneous Truth and Performance Level Estimation (STAPLE) software. Dice and Jaccard indices and other descriptive statistics were computed for the proposed method, as well as the experts' segmentation versus the TS. We also tested the method with the BraTS datasets, which supply expert segmentations. Results and discussion: For 2D segmentation vs. TS, the mean Dice index was 0.90 ± 0.06 (standard deviation), sensitivity was 0.92, and specificity was 0.99. For 3D segmentation vs. TS, the mean Dice index was 0.89 ± 0.06, sensitivity was 0.91, and specificity was 0.99. The automated results are comparable with the experts' manual segmentation results. Conclusions: We present an accurate, robust, efficient, and reproducible segmentation method for pre-operative LGGs.

Original languageEnglish (US)
Article number12
JournalCancer Imaging
Volume15
Issue number1
DOIs
StatePublished - Aug 14 2015

Fingerprint

Glioma
Magnetic Resonance Imaging
Atlases
Neoplasms
Gadolinium
Software
Biomarkers
Brain

ASJC Scopus subject areas

  • Oncology
  • Radiology Nuclear Medicine and imaging
  • Radiological and Ultrasound Technology

Cite this

Semi-automated segmentation of pre-operative low grade gliomas in magnetic resonance imaging. / Akkus, Zeynettin; Sedlar, Jiri; Coufalova, Lucie; Korfiatis, Panagiotis; Kline, Timothy; Warner, Joshua D.; Agrawal, Jay; Erickson, Bradley J.

In: Cancer Imaging, Vol. 15, No. 1, 12, 14.08.2015.

Research output: Contribution to journalArticle

Akkus, Zeynettin ; Sedlar, Jiri ; Coufalova, Lucie ; Korfiatis, Panagiotis ; Kline, Timothy ; Warner, Joshua D. ; Agrawal, Jay ; Erickson, Bradley J. / Semi-automated segmentation of pre-operative low grade gliomas in magnetic resonance imaging. In: Cancer Imaging. 2015 ; Vol. 15, No. 1.
@article{c7fe898935ff4fbc93abe8af2d474f8f,
title = "Semi-automated segmentation of pre-operative low grade gliomas in magnetic resonance imaging",
abstract = "Background: Segmentation of pre-operative low-grade gliomas (LGGs) from magnetic resonance imaging is a crucial step for studying imaging biomarkers. However, segmentation of LGGs is particularly challenging because they rarely enhance after gadolinium administration. Like other gliomas, they have irregular tumor shape, heterogeneous composition, ill-defined tumor boundaries, and limited number of image types. To overcome these challenges we propose a semi-automated segmentation method that relies only on T2-weighted (T2W) and optionally post-contrast T1-weighted (T1W) images. Methods: First, the user draws a region-of-interest (ROI) that completely encloses the tumor and some normal tissue. Second, a normal brain atlas and post-contrast T1W images are registered to T2W images. Third, the posterior probability of each pixel/voxel belonging to normal and abnormal tissues is calculated based on information derived from the atlas and ROI. Finally, geodesic active contours use the probability map of the tumor to shrink the ROI until optimal tumor boundaries are found. This method was validated against the true segmentation (TS) of 30 LGG patients for both 2D (1 slice) and 3D. The TS was obtained from manual segmentations of three experts using the Simultaneous Truth and Performance Level Estimation (STAPLE) software. Dice and Jaccard indices and other descriptive statistics were computed for the proposed method, as well as the experts' segmentation versus the TS. We also tested the method with the BraTS datasets, which supply expert segmentations. Results and discussion: For 2D segmentation vs. TS, the mean Dice index was 0.90 ± 0.06 (standard deviation), sensitivity was 0.92, and specificity was 0.99. For 3D segmentation vs. TS, the mean Dice index was 0.89 ± 0.06, sensitivity was 0.91, and specificity was 0.99. The automated results are comparable with the experts' manual segmentation results. Conclusions: We present an accurate, robust, efficient, and reproducible segmentation method for pre-operative LGGs.",
author = "Zeynettin Akkus and Jiri Sedlar and Lucie Coufalova and Panagiotis Korfiatis and Timothy Kline and Warner, {Joshua D.} and Jay Agrawal and Erickson, {Bradley J}",
year = "2015",
month = "8",
day = "14",
doi = "10.1186/s40644-015-0047-z",
language = "English (US)",
volume = "15",
journal = "Cancer Imaging",
issn = "1740-5025",
publisher = "e-med",
number = "1",

}

TY - JOUR

T1 - Semi-automated segmentation of pre-operative low grade gliomas in magnetic resonance imaging

AU - Akkus, Zeynettin

AU - Sedlar, Jiri

AU - Coufalova, Lucie

AU - Korfiatis, Panagiotis

AU - Kline, Timothy

AU - Warner, Joshua D.

AU - Agrawal, Jay

AU - Erickson, Bradley J

PY - 2015/8/14

Y1 - 2015/8/14

N2 - Background: Segmentation of pre-operative low-grade gliomas (LGGs) from magnetic resonance imaging is a crucial step for studying imaging biomarkers. However, segmentation of LGGs is particularly challenging because they rarely enhance after gadolinium administration. Like other gliomas, they have irregular tumor shape, heterogeneous composition, ill-defined tumor boundaries, and limited number of image types. To overcome these challenges we propose a semi-automated segmentation method that relies only on T2-weighted (T2W) and optionally post-contrast T1-weighted (T1W) images. Methods: First, the user draws a region-of-interest (ROI) that completely encloses the tumor and some normal tissue. Second, a normal brain atlas and post-contrast T1W images are registered to T2W images. Third, the posterior probability of each pixel/voxel belonging to normal and abnormal tissues is calculated based on information derived from the atlas and ROI. Finally, geodesic active contours use the probability map of the tumor to shrink the ROI until optimal tumor boundaries are found. This method was validated against the true segmentation (TS) of 30 LGG patients for both 2D (1 slice) and 3D. The TS was obtained from manual segmentations of three experts using the Simultaneous Truth and Performance Level Estimation (STAPLE) software. Dice and Jaccard indices and other descriptive statistics were computed for the proposed method, as well as the experts' segmentation versus the TS. We also tested the method with the BraTS datasets, which supply expert segmentations. Results and discussion: For 2D segmentation vs. TS, the mean Dice index was 0.90 ± 0.06 (standard deviation), sensitivity was 0.92, and specificity was 0.99. For 3D segmentation vs. TS, the mean Dice index was 0.89 ± 0.06, sensitivity was 0.91, and specificity was 0.99. The automated results are comparable with the experts' manual segmentation results. Conclusions: We present an accurate, robust, efficient, and reproducible segmentation method for pre-operative LGGs.

AB - Background: Segmentation of pre-operative low-grade gliomas (LGGs) from magnetic resonance imaging is a crucial step for studying imaging biomarkers. However, segmentation of LGGs is particularly challenging because they rarely enhance after gadolinium administration. Like other gliomas, they have irregular tumor shape, heterogeneous composition, ill-defined tumor boundaries, and limited number of image types. To overcome these challenges we propose a semi-automated segmentation method that relies only on T2-weighted (T2W) and optionally post-contrast T1-weighted (T1W) images. Methods: First, the user draws a region-of-interest (ROI) that completely encloses the tumor and some normal tissue. Second, a normal brain atlas and post-contrast T1W images are registered to T2W images. Third, the posterior probability of each pixel/voxel belonging to normal and abnormal tissues is calculated based on information derived from the atlas and ROI. Finally, geodesic active contours use the probability map of the tumor to shrink the ROI until optimal tumor boundaries are found. This method was validated against the true segmentation (TS) of 30 LGG patients for both 2D (1 slice) and 3D. The TS was obtained from manual segmentations of three experts using the Simultaneous Truth and Performance Level Estimation (STAPLE) software. Dice and Jaccard indices and other descriptive statistics were computed for the proposed method, as well as the experts' segmentation versus the TS. We also tested the method with the BraTS datasets, which supply expert segmentations. Results and discussion: For 2D segmentation vs. TS, the mean Dice index was 0.90 ± 0.06 (standard deviation), sensitivity was 0.92, and specificity was 0.99. For 3D segmentation vs. TS, the mean Dice index was 0.89 ± 0.06, sensitivity was 0.91, and specificity was 0.99. The automated results are comparable with the experts' manual segmentation results. Conclusions: We present an accurate, robust, efficient, and reproducible segmentation method for pre-operative LGGs.

UR - http://www.scopus.com/inward/record.url?scp=84938915776&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84938915776&partnerID=8YFLogxK

U2 - 10.1186/s40644-015-0047-z

DO - 10.1186/s40644-015-0047-z

M3 - Article

C2 - 26268363

AN - SCOPUS:84938915776

VL - 15

JO - Cancer Imaging

JF - Cancer Imaging

SN - 1740-5025

IS - 1

M1 - 12

ER -