Improved 3D live-wire method with application to 3D CT chest image analysis

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

12 Citations (Scopus)

Abstract

The definition of regions of interests (ROIs), such as suspect cancer nodules or lymph nodes in 3D CT chest images, is often difficult because of the complexity of the phenomena that give rise to them. Manual slice tracing has been used widely for years for such problems, because it is easy to implement and guaranteed to work. But manual slice tracing is extremely time consuming, especially for high-solution 3D images, which may have hundreds of slices, and it is subject to operator biases. Numerous automated image-segmentation methods have been proposed, but they are generally strongly application dependent, and even the "most robust" methods have difficulty in defining complex anatomical ROIs. To address this problem, the semi-automatic interactive paradigm referred to as "live wire" segmentation has been proposed by researchers. In live-wire segmentation, the human operator interactively defines an ROI's boundary guided by an active automated method which suggests what to define. This process in general is far faster, more reproducible and accurate than manual tracing, while, at the same time, permitting the definition of complex ROIs having ill-defined boundaries. We propose a 2D live-wire method employing an improved cost over previous works. In addition, we define a new 3D live-wire formulation that enables rapid definition of 3D ROIs. The method only requires the human operator to consider a few slices in general. Experimental results indicate that the new 2D and 3D live-wire approaches give highly reproducible results (higher than 97% for all ROIs we considered). The 2D live-wire method is roughly at least 3.9 times faster than the manual tracing, while the 3D live-wire method is at least 7.2 times faster than the 2D live-wire approach, or at least 28 times faster than manual tracing. The accuracy of the proposed 2D and 3D live-wire methods was also evaluated by comparing individual segmentation results with corresponding ground truth data. We achieved accuracy rates higher than 95.8% for 2D ROIs and 97.02% for 3D ROIs.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2006
Subtitle of host publicationImage Processing
Volume6144 I
DOIs
StatePublished - Jun 22 2006
EventMedical Imaging 2006: Image Processing - San Diego, CA, United States
Duration: Feb 13 2006Feb 16 2006

Other

OtherMedical Imaging 2006: Image Processing
CountryUnited States
CitySan Diego, CA
Period2/13/062/16/06

Fingerprint

Image analysis
Wire
Image segmentation
Costs

All Science Journal Classification (ASJC) codes

  • Engineering(all)

Cite this

Lu, Kongkuo ; Higgins, William Evan. / Improved 3D live-wire method with application to 3D CT chest image analysis. Medical Imaging 2006: Image Processing. Vol. 6144 I 2006.
@inproceedings{10d9d882f8244adfb3f60bdc00e0c0d9,
title = "Improved 3D live-wire method with application to 3D CT chest image analysis",
abstract = "The definition of regions of interests (ROIs), such as suspect cancer nodules or lymph nodes in 3D CT chest images, is often difficult because of the complexity of the phenomena that give rise to them. Manual slice tracing has been used widely for years for such problems, because it is easy to implement and guaranteed to work. But manual slice tracing is extremely time consuming, especially for high-solution 3D images, which may have hundreds of slices, and it is subject to operator biases. Numerous automated image-segmentation methods have been proposed, but they are generally strongly application dependent, and even the {"}most robust{"} methods have difficulty in defining complex anatomical ROIs. To address this problem, the semi-automatic interactive paradigm referred to as {"}live wire{"} segmentation has been proposed by researchers. In live-wire segmentation, the human operator interactively defines an ROI's boundary guided by an active automated method which suggests what to define. This process in general is far faster, more reproducible and accurate than manual tracing, while, at the same time, permitting the definition of complex ROIs having ill-defined boundaries. We propose a 2D live-wire method employing an improved cost over previous works. In addition, we define a new 3D live-wire formulation that enables rapid definition of 3D ROIs. The method only requires the human operator to consider a few slices in general. Experimental results indicate that the new 2D and 3D live-wire approaches give highly reproducible results (higher than 97{\%} for all ROIs we considered). The 2D live-wire method is roughly at least 3.9 times faster than the manual tracing, while the 3D live-wire method is at least 7.2 times faster than the 2D live-wire approach, or at least 28 times faster than manual tracing. The accuracy of the proposed 2D and 3D live-wire methods was also evaluated by comparing individual segmentation results with corresponding ground truth data. We achieved accuracy rates higher than 95.8{\%} for 2D ROIs and 97.02{\%} for 3D ROIs.",
author = "Kongkuo Lu and Higgins, {William Evan}",
year = "2006",
month = "6",
day = "22",
doi = "10.1117/12.651723",
language = "English (US)",
isbn = "0819464236",
volume = "6144 I",
booktitle = "Medical Imaging 2006",

}

Lu, K & Higgins, WE 2006, Improved 3D live-wire method with application to 3D CT chest image analysis. in Medical Imaging 2006: Image Processing. vol. 6144 I, 61440L, Medical Imaging 2006: Image Processing, San Diego, CA, United States, 2/13/06. https://doi.org/10.1117/12.651723

Improved 3D live-wire method with application to 3D CT chest image analysis. / Lu, Kongkuo; Higgins, William Evan.

Medical Imaging 2006: Image Processing. Vol. 6144 I 2006. 61440L.

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

TY - GEN

T1 - Improved 3D live-wire method with application to 3D CT chest image analysis

AU - Lu, Kongkuo

AU - Higgins, William Evan

PY - 2006/6/22

Y1 - 2006/6/22

N2 - The definition of regions of interests (ROIs), such as suspect cancer nodules or lymph nodes in 3D CT chest images, is often difficult because of the complexity of the phenomena that give rise to them. Manual slice tracing has been used widely for years for such problems, because it is easy to implement and guaranteed to work. But manual slice tracing is extremely time consuming, especially for high-solution 3D images, which may have hundreds of slices, and it is subject to operator biases. Numerous automated image-segmentation methods have been proposed, but they are generally strongly application dependent, and even the "most robust" methods have difficulty in defining complex anatomical ROIs. To address this problem, the semi-automatic interactive paradigm referred to as "live wire" segmentation has been proposed by researchers. In live-wire segmentation, the human operator interactively defines an ROI's boundary guided by an active automated method which suggests what to define. This process in general is far faster, more reproducible and accurate than manual tracing, while, at the same time, permitting the definition of complex ROIs having ill-defined boundaries. We propose a 2D live-wire method employing an improved cost over previous works. In addition, we define a new 3D live-wire formulation that enables rapid definition of 3D ROIs. The method only requires the human operator to consider a few slices in general. Experimental results indicate that the new 2D and 3D live-wire approaches give highly reproducible results (higher than 97% for all ROIs we considered). The 2D live-wire method is roughly at least 3.9 times faster than the manual tracing, while the 3D live-wire method is at least 7.2 times faster than the 2D live-wire approach, or at least 28 times faster than manual tracing. The accuracy of the proposed 2D and 3D live-wire methods was also evaluated by comparing individual segmentation results with corresponding ground truth data. We achieved accuracy rates higher than 95.8% for 2D ROIs and 97.02% for 3D ROIs.

AB - The definition of regions of interests (ROIs), such as suspect cancer nodules or lymph nodes in 3D CT chest images, is often difficult because of the complexity of the phenomena that give rise to them. Manual slice tracing has been used widely for years for such problems, because it is easy to implement and guaranteed to work. But manual slice tracing is extremely time consuming, especially for high-solution 3D images, which may have hundreds of slices, and it is subject to operator biases. Numerous automated image-segmentation methods have been proposed, but they are generally strongly application dependent, and even the "most robust" methods have difficulty in defining complex anatomical ROIs. To address this problem, the semi-automatic interactive paradigm referred to as "live wire" segmentation has been proposed by researchers. In live-wire segmentation, the human operator interactively defines an ROI's boundary guided by an active automated method which suggests what to define. This process in general is far faster, more reproducible and accurate than manual tracing, while, at the same time, permitting the definition of complex ROIs having ill-defined boundaries. We propose a 2D live-wire method employing an improved cost over previous works. In addition, we define a new 3D live-wire formulation that enables rapid definition of 3D ROIs. The method only requires the human operator to consider a few slices in general. Experimental results indicate that the new 2D and 3D live-wire approaches give highly reproducible results (higher than 97% for all ROIs we considered). The 2D live-wire method is roughly at least 3.9 times faster than the manual tracing, while the 3D live-wire method is at least 7.2 times faster than the 2D live-wire approach, or at least 28 times faster than manual tracing. The accuracy of the proposed 2D and 3D live-wire methods was also evaluated by comparing individual segmentation results with corresponding ground truth data. We achieved accuracy rates higher than 95.8% for 2D ROIs and 97.02% for 3D ROIs.

UR - http://www.scopus.com/inward/record.url?scp=33745143615&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33745143615&partnerID=8YFLogxK

U2 - 10.1117/12.651723

DO - 10.1117/12.651723

M3 - Conference contribution

SN - 0819464236

SN - 9780819464231

SN - 0819464236

SN - 9780819464231

VL - 6144 I

BT - Medical Imaging 2006

ER -