Cardiac motion estimation by using high-dimensional features and K-means clustering method

Estanislao Oubel; Alfred O. Hero; Alejandro F. Frangi

doi:10.1117/12.653921

Cardiac motion estimation by using high-dimensional features and K-means clustering method

Estanislao Oubel^*, Alfred O. Hero, Alejandro F. Frangi

^*Corresponding author for this work

Research output: Chapter in Book/Conference proceeding › Conference contribution › peer-review

Abstract

Tagged Magnetic Resonance Imaging (MRI) is currently the reference modality for myocardial motion and strain analysis. Mutual Information (MI) based non rigid registration has proven to be an accurate method to retrieve cardiac motion and overcome many drawbacks present on previous approaches. In a previous work,¹ we used Wavelet-based Attribute Vectors (WAVs) instead of pixel intensity to measure similarity between frames. Since the curse of dimensionality forbids the use of histograms to estimate MI of high dimensional features, k-Nearest Neighbors Graphs (kNNG) were applied to calculate α-MI. Results showed that cardiac motion estimation was feasible with that approach. In this paper, K-Means clustering method is applied to compute MI from the same set of WAVs. The proposed method was applied to four tagging MRI sequences, and the resulting displacements were compared with respect to manual measurements made by two observers. Results show that more accurate motion estimation is obtained with respect to the use of pixel intensity.

Original language	English
Title of host publication	Medical Imaging 2006
Subtitle of host publication	Image Processing
DOIs	https://doi.org/10.1117/12.653921
Publication status	Published - 2006
Event	Medical Imaging 2006: Image Processing - San Diego, CA, United States Duration: 13 Feb 2006 → 16 Feb 2006

Publication series

Name	Progress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume	6144 I
ISSN (Print)	1605-7422

Conference

Conference	Medical Imaging 2006: Image Processing
Country/Territory	United States
City	San Diego, CA
Period	13/02/06 → 16/02/06

Keywords

Cardiac Motion Estimation
High Dimensional Features
K-Means Clustering Method
Non Rigid Registration

Access to Document

10.1117/12.653921

Cite this

@inproceedings{569aa038376a4d83971fcf336c8b9965,

title = "Cardiac motion estimation by using high-dimensional features and K-means clustering method",

abstract = "Tagged Magnetic Resonance Imaging (MRI) is currently the reference modality for myocardial motion and strain analysis. Mutual Information (MI) based non rigid registration has proven to be an accurate method to retrieve cardiac motion and overcome many drawbacks present on previous approaches. In a previous work,1 we used Wavelet-based Attribute Vectors (WAVs) instead of pixel intensity to measure similarity between frames. Since the curse of dimensionality forbids the use of histograms to estimate MI of high dimensional features, k-Nearest Neighbors Graphs (kNNG) were applied to calculate α-MI. Results showed that cardiac motion estimation was feasible with that approach. In this paper, K-Means clustering method is applied to compute MI from the same set of WAVs. The proposed method was applied to four tagging MRI sequences, and the resulting displacements were compared with respect to manual measurements made by two observers. Results show that more accurate motion estimation is obtained with respect to the use of pixel intensity.",

keywords = "Cardiac Motion Estimation, High Dimensional Features, K-Means Clustering Method, Non Rigid Registration",

author = "Estanislao Oubel and Hero, {Alfred O.} and Frangi, {Alejandro F.}",

year = "2006",

doi = "10.1117/12.653921",

language = "English",

isbn = "0819464236",

series = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",

booktitle = "Medical Imaging 2006",

note = "Medical Imaging 2006: Image Processing ; Conference date: 13-02-2006 Through 16-02-2006",

}

Oubel, E, Hero, AO & Frangi, AF 2006, Cardiac motion estimation by using high-dimensional features and K-means clustering method. in Medical Imaging 2006: Image Processing., 614417, Progress in Biomedical Optics and Imaging - Proceedings of SPIE, vol. 6144 I, Medical Imaging 2006: Image Processing, San Diego, CA, United States, 13/02/06. https://doi.org/10.1117/12.653921

Cardiac motion estimation by using high-dimensional features and K-means clustering method. / Oubel, Estanislao; Hero, Alfred O.; Frangi, Alejandro F.
Medical Imaging 2006: Image Processing. 2006. 614417 (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 6144 I).

Research output: Chapter in Book/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Cardiac motion estimation by using high-dimensional features and K-means clustering method

AU - Oubel, Estanislao

AU - Hero, Alfred O.

AU - Frangi, Alejandro F.

PY - 2006

Y1 - 2006

N2 - Tagged Magnetic Resonance Imaging (MRI) is currently the reference modality for myocardial motion and strain analysis. Mutual Information (MI) based non rigid registration has proven to be an accurate method to retrieve cardiac motion and overcome many drawbacks present on previous approaches. In a previous work,1 we used Wavelet-based Attribute Vectors (WAVs) instead of pixel intensity to measure similarity between frames. Since the curse of dimensionality forbids the use of histograms to estimate MI of high dimensional features, k-Nearest Neighbors Graphs (kNNG) were applied to calculate α-MI. Results showed that cardiac motion estimation was feasible with that approach. In this paper, K-Means clustering method is applied to compute MI from the same set of WAVs. The proposed method was applied to four tagging MRI sequences, and the resulting displacements were compared with respect to manual measurements made by two observers. Results show that more accurate motion estimation is obtained with respect to the use of pixel intensity.

AB - Tagged Magnetic Resonance Imaging (MRI) is currently the reference modality for myocardial motion and strain analysis. Mutual Information (MI) based non rigid registration has proven to be an accurate method to retrieve cardiac motion and overcome many drawbacks present on previous approaches. In a previous work,1 we used Wavelet-based Attribute Vectors (WAVs) instead of pixel intensity to measure similarity between frames. Since the curse of dimensionality forbids the use of histograms to estimate MI of high dimensional features, k-Nearest Neighbors Graphs (kNNG) were applied to calculate α-MI. Results showed that cardiac motion estimation was feasible with that approach. In this paper, K-Means clustering method is applied to compute MI from the same set of WAVs. The proposed method was applied to four tagging MRI sequences, and the resulting displacements were compared with respect to manual measurements made by two observers. Results show that more accurate motion estimation is obtained with respect to the use of pixel intensity.

KW - Cardiac Motion Estimation

KW - High Dimensional Features

KW - K-Means Clustering Method

KW - Non Rigid Registration

UR - http://www.scopus.com/inward/record.url?scp=33745122223&partnerID=8YFLogxK

U2 - 10.1117/12.653921

DO - 10.1117/12.653921

M3 - Conference contribution

AN - SCOPUS:33745122223

SN - 0819464236

SN - 9780819464231

SN - 0819464236

SN - 9780819464231

T3 - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

BT - Medical Imaging 2006

T2 - Medical Imaging 2006: Image Processing

Y2 - 13 February 2006 through 16 February 2006

ER -

Cardiac motion estimation by using high-dimensional features and K-means clustering method

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this