Parallel mean shift for interactive volume segmentation

Fangfang Zhou, Ying Zhao, Kwan-Liu Ma

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

15 Citations (Scopus)

Abstract

In this paper we present a parallel dynamic mean shift algorithm based on path transmission for medical volume data segmentation. The algorithm first translates the volume data into a joint position-color feature space subdivided uniformly by bandwidths, and then clusters points in feature space in parallel by iteratively finding its peak point. Over iterations it improves the convergent rate by dynamically updating data points via path transmission and reduces the amount of data points by collapsing overlapping points into one point. The GPU implementation of the algorithm can segment 256x256x256 volume in 6 seconds using an NVIDIA GeForce 8800 GTX card for interactive processing, which is hundreds times faster than its CPU implementation. We also introduce an interactive interface to segment volume data based on this GPU implementation. This interface not only provides the user with the capability to specify segmentation resolution, but also allows the user to operate on the segmented tissues and create desired visualization results.

Original languageEnglish (US)
Title of host publicationMachine Learning in Medical Imaging - First International Workshop, MLMI 2010, Held in Conjunction with MICCAI 2010, Proceedings
Pages67-75
Number of pages9
DOIs
StatePublished - Oct 25 2010
Event1st International Workshop on Machine Learning in Medical Imaging, MLMI 2010, Held in Conjunction with MICCAI 2010 - Beijing, China
Duration: Sep 20 2010Sep 20 2010

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume6357 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other1st International Workshop on Machine Learning in Medical Imaging, MLMI 2010, Held in Conjunction with MICCAI 2010
CountryChina
CityBeijing
Period9/20/109/20/10

Fingerprint

Mean Shift
Segmentation
Feature Space
Interfaces (computer)
Program processors
Visualization
Tissue
Color
Bandwidth
Path
Color Space
Collapsing
Processing
Updating
Overlapping
Iteration
Graphics processing unit

Keywords

  • GPU acceleration
  • Kernel Density Estimation
  • Mean Shift
  • Segmentation
  • Volume Visualization

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Zhou, F., Zhao, Y., & Ma, K-L. (2010). Parallel mean shift for interactive volume segmentation. In Machine Learning in Medical Imaging - First International Workshop, MLMI 2010, Held in Conjunction with MICCAI 2010, Proceedings (pp. 67-75). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6357 LNCS). https://doi.org/10.1007/978-3-642-15948-0_9

Parallel mean shift for interactive volume segmentation. / Zhou, Fangfang; Zhao, Ying; Ma, Kwan-Liu.

Machine Learning in Medical Imaging - First International Workshop, MLMI 2010, Held in Conjunction with MICCAI 2010, Proceedings. 2010. p. 67-75 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6357 LNCS).

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

Zhou, F, Zhao, Y & Ma, K-L 2010, Parallel mean shift for interactive volume segmentation. in Machine Learning in Medical Imaging - First International Workshop, MLMI 2010, Held in Conjunction with MICCAI 2010, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 6357 LNCS, pp. 67-75, 1st International Workshop on Machine Learning in Medical Imaging, MLMI 2010, Held in Conjunction with MICCAI 2010, Beijing, China, 9/20/10. https://doi.org/10.1007/978-3-642-15948-0_9
Zhou F, Zhao Y, Ma K-L. Parallel mean shift for interactive volume segmentation. In Machine Learning in Medical Imaging - First International Workshop, MLMI 2010, Held in Conjunction with MICCAI 2010, Proceedings. 2010. p. 67-75. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-15948-0_9
Zhou, Fangfang ; Zhao, Ying ; Ma, Kwan-Liu. / Parallel mean shift for interactive volume segmentation. Machine Learning in Medical Imaging - First International Workshop, MLMI 2010, Held in Conjunction with MICCAI 2010, Proceedings. 2010. pp. 67-75 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{f202eef991bd4320a6523865de183456,
title = "Parallel mean shift for interactive volume segmentation",
abstract = "In this paper we present a parallel dynamic mean shift algorithm based on path transmission for medical volume data segmentation. The algorithm first translates the volume data into a joint position-color feature space subdivided uniformly by bandwidths, and then clusters points in feature space in parallel by iteratively finding its peak point. Over iterations it improves the convergent rate by dynamically updating data points via path transmission and reduces the amount of data points by collapsing overlapping points into one point. The GPU implementation of the algorithm can segment 256x256x256 volume in 6 seconds using an NVIDIA GeForce 8800 GTX card for interactive processing, which is hundreds times faster than its CPU implementation. We also introduce an interactive interface to segment volume data based on this GPU implementation. This interface not only provides the user with the capability to specify segmentation resolution, but also allows the user to operate on the segmented tissues and create desired visualization results.",
keywords = "GPU acceleration, Kernel Density Estimation, Mean Shift, Segmentation, Volume Visualization",
author = "Fangfang Zhou and Ying Zhao and Kwan-Liu Ma",
year = "2010",
month = "10",
day = "25",
doi = "10.1007/978-3-642-15948-0_9",
language = "English (US)",
isbn = "3642159478",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "67--75",
booktitle = "Machine Learning in Medical Imaging - First International Workshop, MLMI 2010, Held in Conjunction with MICCAI 2010, Proceedings",

}

TY - GEN

T1 - Parallel mean shift for interactive volume segmentation

AU - Zhou, Fangfang

AU - Zhao, Ying

AU - Ma, Kwan-Liu

PY - 2010/10/25

Y1 - 2010/10/25

N2 - In this paper we present a parallel dynamic mean shift algorithm based on path transmission for medical volume data segmentation. The algorithm first translates the volume data into a joint position-color feature space subdivided uniformly by bandwidths, and then clusters points in feature space in parallel by iteratively finding its peak point. Over iterations it improves the convergent rate by dynamically updating data points via path transmission and reduces the amount of data points by collapsing overlapping points into one point. The GPU implementation of the algorithm can segment 256x256x256 volume in 6 seconds using an NVIDIA GeForce 8800 GTX card for interactive processing, which is hundreds times faster than its CPU implementation. We also introduce an interactive interface to segment volume data based on this GPU implementation. This interface not only provides the user with the capability to specify segmentation resolution, but also allows the user to operate on the segmented tissues and create desired visualization results.

AB - In this paper we present a parallel dynamic mean shift algorithm based on path transmission for medical volume data segmentation. The algorithm first translates the volume data into a joint position-color feature space subdivided uniformly by bandwidths, and then clusters points in feature space in parallel by iteratively finding its peak point. Over iterations it improves the convergent rate by dynamically updating data points via path transmission and reduces the amount of data points by collapsing overlapping points into one point. The GPU implementation of the algorithm can segment 256x256x256 volume in 6 seconds using an NVIDIA GeForce 8800 GTX card for interactive processing, which is hundreds times faster than its CPU implementation. We also introduce an interactive interface to segment volume data based on this GPU implementation. This interface not only provides the user with the capability to specify segmentation resolution, but also allows the user to operate on the segmented tissues and create desired visualization results.

KW - GPU acceleration

KW - Kernel Density Estimation

KW - Mean Shift

KW - Segmentation

KW - Volume Visualization

UR - http://www.scopus.com/inward/record.url?scp=77958034510&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77958034510&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-15948-0_9

DO - 10.1007/978-3-642-15948-0_9

M3 - Conference contribution

SN - 3642159478

SN - 9783642159473

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 67

EP - 75

BT - Machine Learning in Medical Imaging - First International Workshop, MLMI 2010, Held in Conjunction with MICCAI 2010, Proceedings

ER -