Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Synthesis of intensity gradient and texture information for efficient three-dimensional segmentation of medical volumes

Synthesis of intensity gradient and texture information for efficient three-dimensional... Abstract. We propose a framework that efficiently employs intensity, gradient, and textural features for three-dimensional (3-D) segmentation of medical (MRI/CT) volumes. Our methodology commences by determining the magnitude of intensity variations across the input volume using a 3-D gradient detection scheme. The resultant gradient volume is utilized in a dynamic volume growing/formation process that is initiated in voxel locations with small gradient magnitudes and is concluded at sites with large gradient magnitudes, yielding a map comprising an initial set of partitions (or subvolumes). This partition map is combined with an entropy-based texture descriptor along with intensity and gradient attributes in a multivariate analysis-based volume merging procedure that fuses subvolumes with similar characteristics to yield a final/refined segmentation output. Additionally, a semiautomated version of the aforestated algorithm that allows a user to interactively segment a desired subvolume of interest as opposed to the entire volume is also discussed. Our approach was tested on several MRI and CT datasets and the results show favorable performance in comparison to the state-of-the-art ITK-SNAP technique. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Medical Imaging SPIE

Synthesis of intensity gradient and texture information for efficient three-dimensional segmentation of medical volumes

Download PDF
18 pages

Loading next page...
 
/lp/spie/synthesis-of-intensity-gradient-and-texture-information-for-efficient-uA9WONpicL

References (41)

Publisher
SPIE
Copyright
© 2015 Society of Photo-Optical Instrumentation Engineers (SPIE)
Subject
Image Processing; Paper
ISSN
2329-4302
eISSN
2329-4310
DOI
10.1117/1.JMI.2.2.024003
pmid
26158098
Publisher site
See Article on Publisher Site

Abstract

Abstract. We propose a framework that efficiently employs intensity, gradient, and textural features for three-dimensional (3-D) segmentation of medical (MRI/CT) volumes. Our methodology commences by determining the magnitude of intensity variations across the input volume using a 3-D gradient detection scheme. The resultant gradient volume is utilized in a dynamic volume growing/formation process that is initiated in voxel locations with small gradient magnitudes and is concluded at sites with large gradient magnitudes, yielding a map comprising an initial set of partitions (or subvolumes). This partition map is combined with an entropy-based texture descriptor along with intensity and gradient attributes in a multivariate analysis-based volume merging procedure that fuses subvolumes with similar characteristics to yield a final/refined segmentation output. Additionally, a semiautomated version of the aforestated algorithm that allows a user to interactively segment a desired subvolume of interest as opposed to the entire volume is also discussed. Our approach was tested on several MRI and CT datasets and the results show favorable performance in comparison to the state-of-the-art ITK-SNAP technique.

Journal

Journal of Medical Imaging – SPIE

Published: Apr 1, 2015

There are no references for this article.