DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Cho, Gyu-Seung | - |
dc.contributor.advisor | 조규성 | - |
dc.contributor.author | Kim, Jin-Sung | - |
dc.contributor.author | 김진성 | - |
dc.date.accessioned | 2011-12-14T08:16:44Z | - |
dc.date.available | 2011-12-14T08:16:44Z | - |
dc.date.issued | 2003 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=180200&flag=dissertation | - |
dc.identifier.uri | http://hdl.handle.net/10203/49464 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 원자력및양자공학과, 2003.2, [ iv, 29 p. ] | - |
dc.description.abstract | We have developed a novel, automatic computer-aided-diagnosis (CAD) program that takes advantage of three-dimensional (3D) volumetric data with multi-slice CT allows us to increase the efficacy of detecting early lung cancer. The efficacy of detecting small nodules increases with high-resolution CT imaging. We have evaluated the effect of slice thickness (ST) and reconstruction interval (RI) on automatic detection of pulmonary nodules in multi-slice CT by a newly developed three-dimensional (3D) morphologic matching (3DMM) CAD algorithm. Clinical spiral chest CT scans obtained in a multi-slice CT were reviewed in five patients with a total of 29 nodules (16>5 mm, 13 of 3-5 mm diameter). From the raw CT projection data, three sets of CT images were reconstructed separately in each patient by selecting ST-RI (mm) at 1-1, 5-1, and 5-5 combinations, resulting in a total of 15 CT data sets. First, 3D lung region was segmented. From this segmented region, non-parenchymal structures were extracted and labeled to generate candidates. Shape features were evaluated along with geometric constraints. For the 5-5 data sets, the program automatically adjusted the 3D shape criteria to compensate ‘low’ 3D connectivity between adjacent slices. 3D region-growing and morphologic matching processes were applied. Finally, the nodules detected by CAD were compared with those identified by a chest radiologist to calculate the sensitivity and false positive rate for each data set. The sensitivity of detecting all nodules ≥3mm by 3DMM CAD was 96.6% ($\frac{28}{29}$) for 1-1, 92.9% ($\frac{26}{28}$) for 5-1, and 87% ($\frac{20}{23}$) for 5-5 data sets. Corresponding false positives per patient (FPP) for nodules ≥ 3mm were 4.6 for 1-1, 9.4 for 5-1, and 24.4 for 5-5 data sets. Detection of pulmonary nodules by our CAD program was highly accurate in high resolution (1mm) CT. The accuracy decreased with an increase in slice thickness but was improved substantially with the use of a small r... | eng |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | reconstruction interval | - |
dc.subject | Slice thickness | - |
dc.subject | Computer Aided Diagnosis | - |
dc.subject | multi-detector Computed Tomography | - |
dc.subject | 다 배열 전산화단층촬영 | - |
dc.subject | 재구성 간격 | - |
dc.subject | 슬라이스 두께 | - |
dc.subject | 자동 검출 알고리즘 | - |
dc.title | Effect of slice thickness and reconstruction interval on automatic detection of pulmonary nodules in multi-slice CT | - |
dc.title.alternative | 다 배열 CT를 이용한 폐 결절 자동 검출 알고리즘에서의 슬라이스 두께와 재구성 간격의 영향에 대한 연구 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 180200/325007 | - |
dc.description.department | 한국과학기술원 : 원자력및양자공학과, | - |
dc.identifier.uid | 020003137 | - |
dc.contributor.localauthor | Kim, Jin-Sung | - |
dc.contributor.localauthor | 김진성 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.