Abstract
One advantage of using instrinsic image properties in medical registration is not necessarily to put fixed markers to identify features which could significantly enlarge the scanning time and cause the patients discomfort. But detecting the same significant landmarks from different image modalities is a high-level image processing task that might prove quite difficult. Using differential geometric features resolves the difficulty of detecting anatomic landmarks. However, operators for differential geometric feature extraction usually generate overwhelming redundant features resulting lack of registration accuracy. In this paper, a second-order differential operator embedded with an eigenimage filter called eigen-Lvv operator is proposed to detect the position of Nasopharyngeal Carcinoma (NPC) in MR images, and then fuse the NPC lesion on the CT image for physicians to comprehend the precise location and tumor range for treatment. The embedded eigenimage filter in this eigen-Lvv operator enhances nasal region, where NPC is located, and suppresses surrounding tissues. Thus, the ridge-like features extracted by the eigen-Lvv operator would focus on the surrounding are of NPC (nasal cavity), aptly avoiding the problem of feature redundancy in most second-order differential operators. A two-stage registration, which consists of a course registration based on geometric contours and a fine registration based on ridge-like anatomical features extracted from eigenimage, is used to reduce the registration time while maintain the registration accuracy. The proposed system has been proven to be effective for NPC detection.
| Original language | English |
|---|---|
| Pages (from-to) | 101-109 |
| Number of pages | 9 |
| Journal | Biomedical Engineering - Applications, Basis and Communications |
| Volume | 12 |
| Issue number | 3 |
| Publication status | Published - 2000 Jun 25 |
Fingerprint
All Science Journal Classification (ASJC) codes
- Biophysics
- Bioengineering
- Biomedical Engineering
Cite this
}
An eigen-support differential operator to fuse CT and MR images for nasopharyngeal carcinoma diagnosis. / Chang, C. Y.; Chung, Pau-Choo; Huang, W. C.; Lai, P. H.
In: Biomedical Engineering - Applications, Basis and Communications, Vol. 12, No. 3, 25.06.2000, p. 101-109.Research output: Contribution to journal › Article
TY - JOUR
T1 - An eigen-support differential operator to fuse CT and MR images for nasopharyngeal carcinoma diagnosis
AU - Chang, C. Y.
AU - Chung, Pau-Choo
AU - Huang, W. C.
AU - Lai, P. H.
PY - 2000/6/25
Y1 - 2000/6/25
N2 - One advantage of using instrinsic image properties in medical registration is not necessarily to put fixed markers to identify features which could significantly enlarge the scanning time and cause the patients discomfort. But detecting the same significant landmarks from different image modalities is a high-level image processing task that might prove quite difficult. Using differential geometric features resolves the difficulty of detecting anatomic landmarks. However, operators for differential geometric feature extraction usually generate overwhelming redundant features resulting lack of registration accuracy. In this paper, a second-order differential operator embedded with an eigenimage filter called eigen-Lvv operator is proposed to detect the position of Nasopharyngeal Carcinoma (NPC) in MR images, and then fuse the NPC lesion on the CT image for physicians to comprehend the precise location and tumor range for treatment. The embedded eigenimage filter in this eigen-Lvv operator enhances nasal region, where NPC is located, and suppresses surrounding tissues. Thus, the ridge-like features extracted by the eigen-Lvv operator would focus on the surrounding are of NPC (nasal cavity), aptly avoiding the problem of feature redundancy in most second-order differential operators. A two-stage registration, which consists of a course registration based on geometric contours and a fine registration based on ridge-like anatomical features extracted from eigenimage, is used to reduce the registration time while maintain the registration accuracy. The proposed system has been proven to be effective for NPC detection.
AB - One advantage of using instrinsic image properties in medical registration is not necessarily to put fixed markers to identify features which could significantly enlarge the scanning time and cause the patients discomfort. But detecting the same significant landmarks from different image modalities is a high-level image processing task that might prove quite difficult. Using differential geometric features resolves the difficulty of detecting anatomic landmarks. However, operators for differential geometric feature extraction usually generate overwhelming redundant features resulting lack of registration accuracy. In this paper, a second-order differential operator embedded with an eigenimage filter called eigen-Lvv operator is proposed to detect the position of Nasopharyngeal Carcinoma (NPC) in MR images, and then fuse the NPC lesion on the CT image for physicians to comprehend the precise location and tumor range for treatment. The embedded eigenimage filter in this eigen-Lvv operator enhances nasal region, where NPC is located, and suppresses surrounding tissues. Thus, the ridge-like features extracted by the eigen-Lvv operator would focus on the surrounding are of NPC (nasal cavity), aptly avoiding the problem of feature redundancy in most second-order differential operators. A two-stage registration, which consists of a course registration based on geometric contours and a fine registration based on ridge-like anatomical features extracted from eigenimage, is used to reduce the registration time while maintain the registration accuracy. The proposed system has been proven to be effective for NPC detection.
UR - http://www.scopus.com/inward/record.url?scp=0034713488&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0034713488&partnerID=8YFLogxK
M3 - Article
AN - SCOPUS:0034713488
VL - 12
SP - 101
EP - 109
JO - Biomedical Engineering - Applications, Basis and Communications
JF - Biomedical Engineering - Applications, Basis and Communications
SN - 1016-2372
IS - 3
ER -