A learning-based CT prostate segmentation method via joint transductive feature selection and regression

Yinghuan Shi, Yaozong Gao, Shu Liao, Daoqiang Zhang, Yang Gao, Dinggang Shen

Research output: Contribution to journalArticle

9 Citations (Scopus)

Abstract

In recent years, there has been a great interest in prostate segmentation, which is an important and challenging task for CT image guided radiotherapy. In this paper, a learning-based segmentation method via joint transductive feature selection and transductive regression is presented, which incorporates the physician's simple manual specification (only taking a few seconds), to aid accurate segmentation, especially for the case with large irregular prostate motion. More specifically, for the current treatment image, experienced physician is first allowed to manually assign the labels for a small subset of prostate and non-prostate voxels, especially in the first and last slices of the prostate regions. Then, the proposed method follows the two step: in prostate-likelihood estimation step, two novel algorithms, tLasso and wLapRLS, will be sequentially employed for transductive feature selection and transductive regression, respectively, aiming to generate the prostate-likelihood map. In multi-atlases based label fusion step, the final segmentation result will be obtained according to the corresponding prostate-likelihood map and the previous images of the same patient. The proposed method has been substantially evaluated on a real prostate CT dataset including 24 patients with 330 CT images, and compared with several state-of-the-art methods. Experimental results show that the proposed method outperforms the state-of-the-arts in terms of higher Dice ratio, higher true positive fraction, and lower centroid distances. Also, the results demonstrate that simple manual specification can help improve the segmentation performance, which is clinically feasible in real practice.

Original languageEnglish
Pages (from-to)317-331
Number of pages15
JournalNeurocomputing
Volume173
DOIs
Publication statusPublished - 2016 Jan 15

Fingerprint

Feature extraction
Labels
Prostate
Joints
Learning
Specifications
Radiotherapy
Fusion reactions
Image-Guided Radiotherapy
Physicians
Atlases

Keywords

  • Feature selection
  • Prostate segmentation
  • Transductive learning

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Science Applications
  • Cognitive Neuroscience

Cite this

A learning-based CT prostate segmentation method via joint transductive feature selection and regression. / Shi, Yinghuan; Gao, Yaozong; Liao, Shu; Zhang, Daoqiang; Gao, Yang; Shen, Dinggang.

In: Neurocomputing, Vol. 173, 15.01.2016, p. 317-331.

Research output: Contribution to journalArticle

Shi, Yinghuan ; Gao, Yaozong ; Liao, Shu ; Zhang, Daoqiang ; Gao, Yang ; Shen, Dinggang. / A learning-based CT prostate segmentation method via joint transductive feature selection and regression. In: Neurocomputing. 2016 ; Vol. 173. pp. 317-331.
@article{777589eed0284be4b6dd4afd184d6294,
title = "A learning-based CT prostate segmentation method via joint transductive feature selection and regression",
abstract = "In recent years, there has been a great interest in prostate segmentation, which is an important and challenging task for CT image guided radiotherapy. In this paper, a learning-based segmentation method via joint transductive feature selection and transductive regression is presented, which incorporates the physician's simple manual specification (only taking a few seconds), to aid accurate segmentation, especially for the case with large irregular prostate motion. More specifically, for the current treatment image, experienced physician is first allowed to manually assign the labels for a small subset of prostate and non-prostate voxels, especially in the first and last slices of the prostate regions. Then, the proposed method follows the two step: in prostate-likelihood estimation step, two novel algorithms, tLasso and wLapRLS, will be sequentially employed for transductive feature selection and transductive regression, respectively, aiming to generate the prostate-likelihood map. In multi-atlases based label fusion step, the final segmentation result will be obtained according to the corresponding prostate-likelihood map and the previous images of the same patient. The proposed method has been substantially evaluated on a real prostate CT dataset including 24 patients with 330 CT images, and compared with several state-of-the-art methods. Experimental results show that the proposed method outperforms the state-of-the-arts in terms of higher Dice ratio, higher true positive fraction, and lower centroid distances. Also, the results demonstrate that simple manual specification can help improve the segmentation performance, which is clinically feasible in real practice.",
keywords = "Feature selection, Prostate segmentation, Transductive learning",
author = "Yinghuan Shi and Yaozong Gao and Shu Liao and Daoqiang Zhang and Yang Gao and Dinggang Shen",
year = "2016",
month = "1",
day = "15",
doi = "10.1016/j.neucom.2014.11.098",
language = "English",
volume = "173",
pages = "317--331",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier",

}

TY - JOUR

T1 - A learning-based CT prostate segmentation method via joint transductive feature selection and regression

AU - Shi, Yinghuan

AU - Gao, Yaozong

AU - Liao, Shu

AU - Zhang, Daoqiang

AU - Gao, Yang

AU - Shen, Dinggang

PY - 2016/1/15

Y1 - 2016/1/15

N2 - In recent years, there has been a great interest in prostate segmentation, which is an important and challenging task for CT image guided radiotherapy. In this paper, a learning-based segmentation method via joint transductive feature selection and transductive regression is presented, which incorporates the physician's simple manual specification (only taking a few seconds), to aid accurate segmentation, especially for the case with large irregular prostate motion. More specifically, for the current treatment image, experienced physician is first allowed to manually assign the labels for a small subset of prostate and non-prostate voxels, especially in the first and last slices of the prostate regions. Then, the proposed method follows the two step: in prostate-likelihood estimation step, two novel algorithms, tLasso and wLapRLS, will be sequentially employed for transductive feature selection and transductive regression, respectively, aiming to generate the prostate-likelihood map. In multi-atlases based label fusion step, the final segmentation result will be obtained according to the corresponding prostate-likelihood map and the previous images of the same patient. The proposed method has been substantially evaluated on a real prostate CT dataset including 24 patients with 330 CT images, and compared with several state-of-the-art methods. Experimental results show that the proposed method outperforms the state-of-the-arts in terms of higher Dice ratio, higher true positive fraction, and lower centroid distances. Also, the results demonstrate that simple manual specification can help improve the segmentation performance, which is clinically feasible in real practice.

AB - In recent years, there has been a great interest in prostate segmentation, which is an important and challenging task for CT image guided radiotherapy. In this paper, a learning-based segmentation method via joint transductive feature selection and transductive regression is presented, which incorporates the physician's simple manual specification (only taking a few seconds), to aid accurate segmentation, especially for the case with large irregular prostate motion. More specifically, for the current treatment image, experienced physician is first allowed to manually assign the labels for a small subset of prostate and non-prostate voxels, especially in the first and last slices of the prostate regions. Then, the proposed method follows the two step: in prostate-likelihood estimation step, two novel algorithms, tLasso and wLapRLS, will be sequentially employed for transductive feature selection and transductive regression, respectively, aiming to generate the prostate-likelihood map. In multi-atlases based label fusion step, the final segmentation result will be obtained according to the corresponding prostate-likelihood map and the previous images of the same patient. The proposed method has been substantially evaluated on a real prostate CT dataset including 24 patients with 330 CT images, and compared with several state-of-the-art methods. Experimental results show that the proposed method outperforms the state-of-the-arts in terms of higher Dice ratio, higher true positive fraction, and lower centroid distances. Also, the results demonstrate that simple manual specification can help improve the segmentation performance, which is clinically feasible in real practice.

KW - Feature selection

KW - Prostate segmentation

KW - Transductive learning

UR - http://www.scopus.com/inward/record.url?scp=84948665052&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84948665052&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2014.11.098

DO - 10.1016/j.neucom.2014.11.098

M3 - Article

AN - SCOPUS:84948665052

VL - 173

SP - 317

EP - 331

JO - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

ER -