Liver tumour segmentation using contrast-enhanced multi-detector CT data: Performance benchmarking of three semiautomated methods

Jia Yin Zhou, Damon W K Wong, Feng Ding, Sudhakar K Venkatesh, Qi Tian, Ying Yi Qi, Wei Xiong, Jimmy J. Liu, Wee Kheng Leow

Research output: Contribution to journalArticle

30 Citations (Scopus)

Abstract

Objective: Automatic tumour segmentation and volumetry is useful in cancer staging and treatment outcome assessment. This paper presents a performance benchmarking study on liver tumour segmentation for three semiautomatic algorithms: 2D region growing with knowledge-based constraints (A1), 2D voxel classification with propagational learning (A2) and Bayesian rule-based 3D region growing (A3). Methods: CT data from 30 patients were studied, and 47 liver tumours were isolated and manually segmented by experts to obtain the reference standard. Four datasets with ten tumours were used for algorithm training and the remaining 37 tumours for testing. Three evaluation metrics, relative absolute volume difference (RAVD), volumetric overlap error (VOE) and average symmetric surface distance (ASSD), were computed based on computerised and reference segmentations. Results: A1, A2 and A3 obtained mean/median RAVD scores of 17.93/10.53%, 17.92/9.61% and 34.74/28.75%, mean/median VOEs of 30.47/26.79%, 25.70/22.64% and 39.95/38.54%, and mean/median ASSDs of 2.05/1.41 mm, 1.57/1.15 mm and 4.12/3.41 mm, respectively. For each metric, we obtained significantly lower values of A1 and A2 than A3 (P<0.01), suggesting that A1 and A2 outperformed A3. Conclusions: Compared with the reference standard, the overall performance of A1 and A2 is promising. Further development and validation is necessary before reliable tumour segmentation and volumetry can be widely used clinically.

Original languageEnglish (US)
Pages (from-to)1738-1748
Number of pages11
JournalEuropean Radiology
Volume20
Issue number7
DOIs
StatePublished - Jul 2010
Externally publishedYes

Fingerprint

Benchmarking
varespladib methyl
Liver
Neoplasms
Neoplasm Staging
Outcome Assessment (Health Care)
Learning

Keywords

  • Computed tomography (CT)
  • Image segmentation
  • Liver tumour
  • Performance benchmarking
  • Tumour volumetry

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging

Cite this

Liver tumour segmentation using contrast-enhanced multi-detector CT data : Performance benchmarking of three semiautomated methods. / Zhou, Jia Yin; Wong, Damon W K; Ding, Feng; Venkatesh, Sudhakar K; Tian, Qi; Qi, Ying Yi; Xiong, Wei; Liu, Jimmy J.; Leow, Wee Kheng.

In: European Radiology, Vol. 20, No. 7, 07.2010, p. 1738-1748.

Research output: Contribution to journalArticle

Zhou, Jia Yin ; Wong, Damon W K ; Ding, Feng ; Venkatesh, Sudhakar K ; Tian, Qi ; Qi, Ying Yi ; Xiong, Wei ; Liu, Jimmy J. ; Leow, Wee Kheng. / Liver tumour segmentation using contrast-enhanced multi-detector CT data : Performance benchmarking of three semiautomated methods. In: European Radiology. 2010 ; Vol. 20, No. 7. pp. 1738-1748.
@article{4e9320eb01bd4f0c9a1f3ac7ec35eb0a,
title = "Liver tumour segmentation using contrast-enhanced multi-detector CT data: Performance benchmarking of three semiautomated methods",
abstract = "Objective: Automatic tumour segmentation and volumetry is useful in cancer staging and treatment outcome assessment. This paper presents a performance benchmarking study on liver tumour segmentation for three semiautomatic algorithms: 2D region growing with knowledge-based constraints (A1), 2D voxel classification with propagational learning (A2) and Bayesian rule-based 3D region growing (A3). Methods: CT data from 30 patients were studied, and 47 liver tumours were isolated and manually segmented by experts to obtain the reference standard. Four datasets with ten tumours were used for algorithm training and the remaining 37 tumours for testing. Three evaluation metrics, relative absolute volume difference (RAVD), volumetric overlap error (VOE) and average symmetric surface distance (ASSD), were computed based on computerised and reference segmentations. Results: A1, A2 and A3 obtained mean/median RAVD scores of 17.93/10.53{\%}, 17.92/9.61{\%} and 34.74/28.75{\%}, mean/median VOEs of 30.47/26.79{\%}, 25.70/22.64{\%} and 39.95/38.54{\%}, and mean/median ASSDs of 2.05/1.41 mm, 1.57/1.15 mm and 4.12/3.41 mm, respectively. For each metric, we obtained significantly lower values of A1 and A2 than A3 (P<0.01), suggesting that A1 and A2 outperformed A3. Conclusions: Compared with the reference standard, the overall performance of A1 and A2 is promising. Further development and validation is necessary before reliable tumour segmentation and volumetry can be widely used clinically.",
keywords = "Computed tomography (CT), Image segmentation, Liver tumour, Performance benchmarking, Tumour volumetry",
author = "Zhou, {Jia Yin} and Wong, {Damon W K} and Feng Ding and Venkatesh, {Sudhakar K} and Qi Tian and Qi, {Ying Yi} and Wei Xiong and Liu, {Jimmy J.} and Leow, {Wee Kheng}",
year = "2010",
month = "7",
doi = "10.1007/s00330-010-1712-z",
language = "English (US)",
volume = "20",
pages = "1738--1748",
journal = "European Radiology",
issn = "0938-7994",
publisher = "Springer Verlag",
number = "7",

}

TY - JOUR

T1 - Liver tumour segmentation using contrast-enhanced multi-detector CT data

T2 - Performance benchmarking of three semiautomated methods

AU - Zhou, Jia Yin

AU - Wong, Damon W K

AU - Ding, Feng

AU - Venkatesh, Sudhakar K

AU - Tian, Qi

AU - Qi, Ying Yi

AU - Xiong, Wei

AU - Liu, Jimmy J.

AU - Leow, Wee Kheng

PY - 2010/7

Y1 - 2010/7

N2 - Objective: Automatic tumour segmentation and volumetry is useful in cancer staging and treatment outcome assessment. This paper presents a performance benchmarking study on liver tumour segmentation for three semiautomatic algorithms: 2D region growing with knowledge-based constraints (A1), 2D voxel classification with propagational learning (A2) and Bayesian rule-based 3D region growing (A3). Methods: CT data from 30 patients were studied, and 47 liver tumours were isolated and manually segmented by experts to obtain the reference standard. Four datasets with ten tumours were used for algorithm training and the remaining 37 tumours for testing. Three evaluation metrics, relative absolute volume difference (RAVD), volumetric overlap error (VOE) and average symmetric surface distance (ASSD), were computed based on computerised and reference segmentations. Results: A1, A2 and A3 obtained mean/median RAVD scores of 17.93/10.53%, 17.92/9.61% and 34.74/28.75%, mean/median VOEs of 30.47/26.79%, 25.70/22.64% and 39.95/38.54%, and mean/median ASSDs of 2.05/1.41 mm, 1.57/1.15 mm and 4.12/3.41 mm, respectively. For each metric, we obtained significantly lower values of A1 and A2 than A3 (P<0.01), suggesting that A1 and A2 outperformed A3. Conclusions: Compared with the reference standard, the overall performance of A1 and A2 is promising. Further development and validation is necessary before reliable tumour segmentation and volumetry can be widely used clinically.

AB - Objective: Automatic tumour segmentation and volumetry is useful in cancer staging and treatment outcome assessment. This paper presents a performance benchmarking study on liver tumour segmentation for three semiautomatic algorithms: 2D region growing with knowledge-based constraints (A1), 2D voxel classification with propagational learning (A2) and Bayesian rule-based 3D region growing (A3). Methods: CT data from 30 patients were studied, and 47 liver tumours were isolated and manually segmented by experts to obtain the reference standard. Four datasets with ten tumours were used for algorithm training and the remaining 37 tumours for testing. Three evaluation metrics, relative absolute volume difference (RAVD), volumetric overlap error (VOE) and average symmetric surface distance (ASSD), were computed based on computerised and reference segmentations. Results: A1, A2 and A3 obtained mean/median RAVD scores of 17.93/10.53%, 17.92/9.61% and 34.74/28.75%, mean/median VOEs of 30.47/26.79%, 25.70/22.64% and 39.95/38.54%, and mean/median ASSDs of 2.05/1.41 mm, 1.57/1.15 mm and 4.12/3.41 mm, respectively. For each metric, we obtained significantly lower values of A1 and A2 than A3 (P<0.01), suggesting that A1 and A2 outperformed A3. Conclusions: Compared with the reference standard, the overall performance of A1 and A2 is promising. Further development and validation is necessary before reliable tumour segmentation and volumetry can be widely used clinically.

KW - Computed tomography (CT)

KW - Image segmentation

KW - Liver tumour

KW - Performance benchmarking

KW - Tumour volumetry

UR - http://www.scopus.com/inward/record.url?scp=77954819248&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77954819248&partnerID=8YFLogxK

U2 - 10.1007/s00330-010-1712-z

DO - 10.1007/s00330-010-1712-z

M3 - Article

C2 - 20157817

AN - SCOPUS:77954819248

VL - 20

SP - 1738

EP - 1748

JO - European Radiology

JF - European Radiology

SN - 0938-7994

IS - 7

ER -