2D MRI brain segmentation by using feasibility constraints

Valentina Pedoia, Elisabetta Binaghi, Sergio Balbi, Alessandro De Benedictis, Emanuele Monti, Renzo Minotto

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

Abstract

Perform a robust MRI brain segmentation is an hard task, specially in presence of tumor that affect the normal brain tissue texture and intensity features. The brain tumors can take various shape, position and intensity level. For these reasons perform an accurate and general brain tumor segmentation is very difficult. The brain segmentation is an essential pre processing task for develop a good tumor detection and segmentation algorithm. In this paper we propose a new 2D brain segmentation algorithm that use graph searching principles. The aim of our work is to translate the segmentation problem in a constrained optimization easily solvable with a graph. To achieve this goal the brain boundary must be described by a feasible function and a setof constrains easily describable through the connections of a graph. Trough a preprocessing phase that "unwrup" the image in polar coordinates the brain boundary is a feasible function that can be found looking for the minimal path of a weighted graph with the dynamic programming. The brain segmentation algorithm proposed in this paper is a fully automatic and nonsupervised method and don't consider a priori condition. The algorithm is validated on a set of volumetric FLAIR MR image, it appears robust in the presence of tumors different in terms of shape and position.

Original languageEnglish
Title of host publicationComputational Vision and Medical Image Processing, Proceedings of VipIMAGE 2011 - 3rd ECCOMAS Thematic Conference on Computational Vision and Medical Image Processing
Pages251-256
Number of pages6
Publication statusPublished - 2012
Event3rd ECCOMAS Thematic Conference on Computational Vision and Medical Image Processing, VipIMAGE 2011 - Olhao, Algarve, Portugal
Duration: Oct 12 2011Oct 14 2011

Other

Other3rd ECCOMAS Thematic Conference on Computational Vision and Medical Image Processing, VipIMAGE 2011
CountryPortugal
CityOlhao, Algarve
Period10/12/1110/14/11

Fingerprint

Magnetic resonance imaging
Brain
Tumors
Constrained optimization
Dynamic programming
Textures
Tissue
Processing

Keywords

  • Brain segmentation
  • Constraints
  • Feasibility
  • Graph optimization
  • MRI
  • Polar conversion

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Biomedical Engineering

Cite this

Pedoia, V., Binaghi, E., Balbi, S., De Benedictis, A., Monti, E., & Minotto, R. (2012). 2D MRI brain segmentation by using feasibility constraints. In Computational Vision and Medical Image Processing, Proceedings of VipIMAGE 2011 - 3rd ECCOMAS Thematic Conference on Computational Vision and Medical Image Processing (pp. 251-256)

2D MRI brain segmentation by using feasibility constraints. / Pedoia, Valentina; Binaghi, Elisabetta; Balbi, Sergio; De Benedictis, Alessandro; Monti, Emanuele; Minotto, Renzo.

Computational Vision and Medical Image Processing, Proceedings of VipIMAGE 2011 - 3rd ECCOMAS Thematic Conference on Computational Vision and Medical Image Processing. 2012. p. 251-256.

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

Pedoia, V, Binaghi, E, Balbi, S, De Benedictis, A, Monti, E & Minotto, R 2012, 2D MRI brain segmentation by using feasibility constraints. in Computational Vision and Medical Image Processing, Proceedings of VipIMAGE 2011 - 3rd ECCOMAS Thematic Conference on Computational Vision and Medical Image Processing. pp. 251-256, 3rd ECCOMAS Thematic Conference on Computational Vision and Medical Image Processing, VipIMAGE 2011, Olhao, Algarve, Portugal, 10/12/11.
Pedoia V, Binaghi E, Balbi S, De Benedictis A, Monti E, Minotto R. 2D MRI brain segmentation by using feasibility constraints. In Computational Vision and Medical Image Processing, Proceedings of VipIMAGE 2011 - 3rd ECCOMAS Thematic Conference on Computational Vision and Medical Image Processing. 2012. p. 251-256
Pedoia, Valentina ; Binaghi, Elisabetta ; Balbi, Sergio ; De Benedictis, Alessandro ; Monti, Emanuele ; Minotto, Renzo. / 2D MRI brain segmentation by using feasibility constraints. Computational Vision and Medical Image Processing, Proceedings of VipIMAGE 2011 - 3rd ECCOMAS Thematic Conference on Computational Vision and Medical Image Processing. 2012. pp. 251-256
@inproceedings{6eb26df05e0d42bd8db273de6d439bcb,
title = "2D MRI brain segmentation by using feasibility constraints",
abstract = "Perform a robust MRI brain segmentation is an hard task, specially in presence of tumor that affect the normal brain tissue texture and intensity features. The brain tumors can take various shape, position and intensity level. For these reasons perform an accurate and general brain tumor segmentation is very difficult. The brain segmentation is an essential pre processing task for develop a good tumor detection and segmentation algorithm. In this paper we propose a new 2D brain segmentation algorithm that use graph searching principles. The aim of our work is to translate the segmentation problem in a constrained optimization easily solvable with a graph. To achieve this goal the brain boundary must be described by a feasible function and a setof constrains easily describable through the connections of a graph. Trough a preprocessing phase that {"}unwrup{"} the image in polar coordinates the brain boundary is a feasible function that can be found looking for the minimal path of a weighted graph with the dynamic programming. The brain segmentation algorithm proposed in this paper is a fully automatic and nonsupervised method and don't consider a priori condition. The algorithm is validated on a set of volumetric FLAIR MR image, it appears robust in the presence of tumors different in terms of shape and position.",
keywords = "Brain segmentation, Constraints, Feasibility, Graph optimization, MRI, Polar conversion",
author = "Valentina Pedoia and Elisabetta Binaghi and Sergio Balbi and {De Benedictis}, Alessandro and Emanuele Monti and Renzo Minotto",
year = "2012",
language = "English",
isbn = "9780415683951",
pages = "251--256",
booktitle = "Computational Vision and Medical Image Processing, Proceedings of VipIMAGE 2011 - 3rd ECCOMAS Thematic Conference on Computational Vision and Medical Image Processing",

}

TY - GEN

T1 - 2D MRI brain segmentation by using feasibility constraints

AU - Pedoia, Valentina

AU - Binaghi, Elisabetta

AU - Balbi, Sergio

AU - De Benedictis, Alessandro

AU - Monti, Emanuele

AU - Minotto, Renzo

PY - 2012

Y1 - 2012

N2 - Perform a robust MRI brain segmentation is an hard task, specially in presence of tumor that affect the normal brain tissue texture and intensity features. The brain tumors can take various shape, position and intensity level. For these reasons perform an accurate and general brain tumor segmentation is very difficult. The brain segmentation is an essential pre processing task for develop a good tumor detection and segmentation algorithm. In this paper we propose a new 2D brain segmentation algorithm that use graph searching principles. The aim of our work is to translate the segmentation problem in a constrained optimization easily solvable with a graph. To achieve this goal the brain boundary must be described by a feasible function and a setof constrains easily describable through the connections of a graph. Trough a preprocessing phase that "unwrup" the image in polar coordinates the brain boundary is a feasible function that can be found looking for the minimal path of a weighted graph with the dynamic programming. The brain segmentation algorithm proposed in this paper is a fully automatic and nonsupervised method and don't consider a priori condition. The algorithm is validated on a set of volumetric FLAIR MR image, it appears robust in the presence of tumors different in terms of shape and position.

AB - Perform a robust MRI brain segmentation is an hard task, specially in presence of tumor that affect the normal brain tissue texture and intensity features. The brain tumors can take various shape, position and intensity level. For these reasons perform an accurate and general brain tumor segmentation is very difficult. The brain segmentation is an essential pre processing task for develop a good tumor detection and segmentation algorithm. In this paper we propose a new 2D brain segmentation algorithm that use graph searching principles. The aim of our work is to translate the segmentation problem in a constrained optimization easily solvable with a graph. To achieve this goal the brain boundary must be described by a feasible function and a setof constrains easily describable through the connections of a graph. Trough a preprocessing phase that "unwrup" the image in polar coordinates the brain boundary is a feasible function that can be found looking for the minimal path of a weighted graph with the dynamic programming. The brain segmentation algorithm proposed in this paper is a fully automatic and nonsupervised method and don't consider a priori condition. The algorithm is validated on a set of volumetric FLAIR MR image, it appears robust in the presence of tumors different in terms of shape and position.

KW - Brain segmentation

KW - Constraints

KW - Feasibility

KW - Graph optimization

KW - MRI

KW - Polar conversion

UR - http://www.scopus.com/inward/record.url?scp=84856726756&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84856726756&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84856726756

SN - 9780415683951

SP - 251

EP - 256

BT - Computational Vision and Medical Image Processing, Proceedings of VipIMAGE 2011 - 3rd ECCOMAS Thematic Conference on Computational Vision and Medical Image Processing

ER -