Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2005 Oct;33(10):1439-48.
doi: 10.1007/s10439-005-5888-3.

Implementation of high-dimensional feature map for segmentation of MR images

Renjie He  1 , Balasrinivasa Rao Sajja, Ponnada A Narayana
Affiliations

Implementation of high-dimensional feature map for segmentation of MR images

Renjie He et al. Ann Biomed Eng. 2005 Oct.

Abstract

A method that considerably reduces the computational and memory complexities associated with the generation of high-dimensional (> or =3) feature maps for image segmentation is described. The method is based on the K-nearest neighbor (KNN) classification and consists of two parts: preprocessing of feature space and fast KNN. This technique is implemented on a PC and applied for generating 3D and 4D feature maps for segmenting MR brain images of multiple sclerosis patients.

PubMed Disclaimer

Figures

FIGURE 1
FIGURE 1
Schematic representation of the 2D feature space. (a) Whole feature space inside the solid line, (b) division of feature space into subspaces, (c) seed prototypes inside the overlapping dashed line regions that affect the boundaries of a subspace, (d) dashed square representing the region that needs to be included for classification of feature vectors inside one subspace. The solid dots represent the feature vectors.
FIGURE 2
FIGURE 2
(a) Procedures for reducing the number of feature vectors for KNN classification in one subspace. Seed prototypes are shown as solid small circles. (b) The unclassified feature vectors in the entire feature space are restricted to the interfaces between the classes.
FIGURE 3
FIGURE 3
Regions showing the seed prototypes that need to be taken into consideration.
FIGURE 4
FIGURE 4
Registration of FLAIR and short TE FSE images, (a) FLAIR image before registration, (b) FLAIR image after registration, (c) target image (short echo FSE, also referred to as the proton density (PD)) .
FIGURE 5
FIGURE 5
FSE and FLAIR images prior to (top row) and following (bottom row) bias correction.
FIGURE 6
FIGURE 6
FSE and FLAIR images prior to (top row) and following stripping (bottom row).
FIGURE 7
FIGURE 7
FSE and FLAIR images prior to (top row) and following (bottom row) filtration.
FIGURE 8
FIGURE 8
FSE and FLAIR images prior to (top row) and following (bottom row) histogram standardization.
FIGURE 9
FIGURE 9
Visualization tool for 2D application. First column is image viewer for seeds pick up, second column is feature map and seeds viewer, and third column is seeds editor.
FIGURE 10
FIGURE 10
Visualization tool for 3D application. (a) Image viewer, analyzer and seeds pick up, (b) seeds viewer and editor, (c) feature map viewer.
FIGURE 11
FIGURE 11
(a)-(d): Four AFFIRMATIVE images. (e): segmented image using 4D feature map.
FIGURE 12
FIGURE 12
Sections of 4D feature maps of one subspace, (a) is generated before applying the fast KNN algorithm, and (b) is after applying the fast KNN algorithm.
FIGURE 13
FIGURE 13
Quantitative analyses of 4D segmentation of AFFIRMATIVE images. The symbols diamond, square, triangle, and cross represent gray matter, white matter, CSF, and lesion respectively.
FIGURE 14
FIGURE 14
Segmentation on FLAIR+FSE.
FIGURE 15
FIGURE 15
Quantitative analyses of 3D segmentation of FSE+FLAIR images. The symbols diamond, square, triangle, and cross represent gray matter, white matter, CSF, and lesion respectively.
See this image and copyright information in PMC

References

    1. Amato U, Larobina M, Antoniadis A, Alfano B. Segmentation of magnetic resonance brain images through discriminant analysis. J Neurosci Methods. 2003;131(1–2):65–74. - PubMed
    1. Anbeek P, Vincken KL, van Osch MJP, Bisschops RHC, Grond J. Probabilistic segmentation of white matter lesions in MR imaging. NeuroImage. 2004;21:1037–1044. - PubMed
    1. Ashburner, J.. Another MRI Bias Correction Approach. The 8th International Conference on Functional Mapping of the Human Brain, Sendai, Japan, June 2–6, 2002. Available on CD-Rom, also appeared in NeuroImage, 16(Suppl. No. 2).
    1. Bedell BJ, Narayana PA. Volumetric analysis of white matter, gray matter, and CSF using fractional volume analysis. Magn Reson Med. 1998;39:961–969. - PubMed
    1. Bedell BJ, Narayana PA, Wolinsky JS. A dual approach for minimizing false lesions classifications on magnetic resonance images. Magn Reson Med. 1997;37:94–102. - PubMed

Publication types

MeSH terms