你的查詢指令已經更改... 已嘗試: (diagnostic imaging digital techniques congr) 無記錄查獲..... 已嘗試: (diagnostic and imaging and digital and techniques and congr) 無記錄查獲..... 已嘗試: (diagnostic or imaging or digital or techniques or congr)
32000 結果查獲. 排序依照 相關性 .
Record:   Prev Next
作者 Yuan, Yading
書名 Correlative analysis of breast lesions on full-field digital mammography and magnetic resonance imaging
國際標準書號 9781124049533
book jacket
說明 154 p
附註 Source: Dissertation Abstracts International, Volume: 71-07, Section: B, page: 4319
Adviser: Maryellen L. Giger
Thesis (Ph.D.)--The University of Chicago, 2010
Multi-modality imaging techniques are increasingly being applied in clinical practice to improve the accuracy with which breast cancer can be diagnosed. However, interpreting images from different modalities is not trivial as different images of the same lesion may exhibit different physical lesion attributes, and currently the various image modality acquisitions are performed under different breast positioning protocols
The general objective of this research is to investigate computerized correlative feature analysis (CFA) methods for integrating information from full-field digital mammographic (FFDM) images and dynamic contrast-enhanced magnetic resonance (DCE-MR) images by taking advantage of the information from different imaging modalities, and thus improving the diagnostic ability of computer-aided diagnosis (CADx) in breast cancer workup. The main hypothesis to be tested is that by incorporating correlative feature analysis in CADx, one can achieve an accurate and efficient discrimination between corresponding and non-corresponding lesion pairs, and subsequently improve performance in the estimation of computer-estimated probabilities of malignancy
The main contributions of this research work are summarized as follows. (1) A novel active-contour model based algorithm was developed for lesion segmentation on mammograms. This new algorithm yielded a statistically improved segmentation performance as compared to previously developed methods: a region-growing method and a radial gradient index (RGI) based method. (2) A computerized feature-based, supervised-learning driven CFA method was investigated to identify corresponding lesions in different mammographic views. The performance obtained by combining multiple features was found to be statistically better than the use of a distance feature alone, and robust across different mammographic view combinations. (3) A multi-modality CADx method that automatically selects and combines discriminative information from both mammography and DCE-MR imaging was studied, and yielded a statistically improved diagnostic performance as compared to the use of single-modality CADx. (4) The CFA method was successfully generalized to the task of differentiating between corresponding and non-corresponding lesions seen in mammographic images and DCE-MR images. Furthermore, multi-modality CADx, which incorporated CFA, was found to potentially provide improved diagnostic accuracy as compared to both single-modality CADx and to multi-modality CADx that erroneously includes non-corresponding lesion pairs
The results affirm the main hypothesis of this research work. With the increasing role of multi-modality imaging techniques in the clinical evaluation, computerized analysis, integration, and interpretation of the image data become more essential for breast cancer diagnosis. The significance of this research is that it provides an automated, effective and efficient scheme that has potential to help radiologists in achieving an improved correlation and characterization of breast lesions with multi-modality imaging techniques
School code: 0330
Host Item Dissertation Abstracts International 71-07B
主題 Health Sciences, Radiology
Physics, Radiation
0574
0756
Alt Author The University of Chicago. Medical Physics
Record:   Prev Next