Abstract
Dedicated breast CT (bCT) is an emerging technology that produces 3D reconstructed images of the breast, thus allowing radiologists to detect and evaluate breast lesions in 3D. In previous work, we have developed an algorithm that combines radial gradient index (RGI) segmentation and a modified level set model for segmentation of lesions in contrast-enhanced bCT images; yielding an average overlap ratio (OR) of 0.69, which is higher than 0.4, the overlap ratio that is generally deemed "acceptable". In this study, this segmentation algorithm, with the same parameter settings, was applied to the corresponding non-contrastenhanced bCT images. The results show that the OR obtained on non-contrast images was 0.62, with the segmented lesion volumes tending to be slightly smaller as compared with those obtained on the corresponding contrast-enhanced images. These results imply that while use of contrast improves segmentation performance, the increase may not be significant, and thus, the role of noncontrast-enhanced breast CT should be further investigated.
| Original language | English (US) |
|---|---|
| Title of host publication | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
| Pages | 697-704 |
| Number of pages | 8 |
| Volume | 7361 LNCS |
| DOIs | |
| State | Published - 2012 |
| Event | 11th International Workshop on Breast Imaging, IWDM 2012 - Philadelphia, PA, United States Duration: Jul 8 2012 → Jul 11 2012 |
Publication series
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Volume | 7361 LNCS |
| ISSN (Print) | 03029743 |
| ISSN (Electronic) | 16113349 |
Other
| Other | 11th International Workshop on Breast Imaging, IWDM 2012 |
|---|---|
| Country | United States |
| City | Philadelphia, PA |
| Period | 7/8/12 → 7/11/12 |
Fingerprint
ASJC Scopus subject areas
- Computer Science(all)
- Theoretical Computer Science
Cite this
Level set breast mass segmentation in contrast-enhanced and non-contrast-enhanced breast CT. / Kuo, Hsien Chi; Giger, Maryellen L.; Reiser, Ingrid; Boone, John M; Lindfors, Karen K; Yang, Kai; Edwards, Alexandra.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 7361 LNCS 2012. p. 697-704 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7361 LNCS).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Level set breast mass segmentation in contrast-enhanced and non-contrast-enhanced breast CT
AU - Kuo, Hsien Chi
AU - Giger, Maryellen L.
AU - Reiser, Ingrid
AU - Boone, John M
AU - Lindfors, Karen K
AU - Yang, Kai
AU - Edwards, Alexandra
PY - 2012
Y1 - 2012
N2 - Dedicated breast CT (bCT) is an emerging technology that produces 3D reconstructed images of the breast, thus allowing radiologists to detect and evaluate breast lesions in 3D. In previous work, we have developed an algorithm that combines radial gradient index (RGI) segmentation and a modified level set model for segmentation of lesions in contrast-enhanced bCT images; yielding an average overlap ratio (OR) of 0.69, which is higher than 0.4, the overlap ratio that is generally deemed "acceptable". In this study, this segmentation algorithm, with the same parameter settings, was applied to the corresponding non-contrastenhanced bCT images. The results show that the OR obtained on non-contrast images was 0.62, with the segmented lesion volumes tending to be slightly smaller as compared with those obtained on the corresponding contrast-enhanced images. These results imply that while use of contrast improves segmentation performance, the increase may not be significant, and thus, the role of noncontrast-enhanced breast CT should be further investigated.
AB - Dedicated breast CT (bCT) is an emerging technology that produces 3D reconstructed images of the breast, thus allowing radiologists to detect and evaluate breast lesions in 3D. In previous work, we have developed an algorithm that combines radial gradient index (RGI) segmentation and a modified level set model for segmentation of lesions in contrast-enhanced bCT images; yielding an average overlap ratio (OR) of 0.69, which is higher than 0.4, the overlap ratio that is generally deemed "acceptable". In this study, this segmentation algorithm, with the same parameter settings, was applied to the corresponding non-contrastenhanced bCT images. The results show that the OR obtained on non-contrast images was 0.62, with the segmented lesion volumes tending to be slightly smaller as compared with those obtained on the corresponding contrast-enhanced images. These results imply that while use of contrast improves segmentation performance, the increase may not be significant, and thus, the role of noncontrast-enhanced breast CT should be further investigated.
UR - http://www.scopus.com/inward/record.url?scp=84864857095&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84864857095&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-31271-7_90
DO - 10.1007/978-3-642-31271-7_90
M3 - Conference contribution
AN - SCOPUS:84864857095
SN - 9783642312700
VL - 7361 LNCS
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 697
EP - 704
BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
ER -