| Summary: | 博士 === 國立中正大學 === 資訊工程所 === 94 === Recently, image registration has been applied popularly into medical imaging and several superior results have also been demonstrated. Image registration technique refers to a process of determining a spatial transform mapping on one image into another. These applications involving image registration include registering images of the same subject from different modalities and aligning temporal sequences of images to compensate for motion of the subject between scans. Besides, image registration can be divided into rigid and non-rigid registration, respectively. In this thesis, we will present several studies respectively related to rigid and non-rigid image registration techniques on breast ultrasound.
In the study for rigid registration, we proposed a newly developed algorithm that can integrate the useful information from the separately scanned ultrasound images and stitch these images into a full-view breast image using the rigid-body image registration. The produced full-view images can reveal the breast anatomy and assists physicians to reduce extra manual adjustment. Moreover, these full-view images also can be applied into the computer-aided diagnosis (CAD) system to detect the suspicious tumor regions. Second, we utilized non-rigid image registration modality on the analysis of tumor deformation. Because the shape of a tumor may be altered due to the stress caused by the ultrasound probe, the non-rigid image registration can be utilized to analyze the variation of tumor between pre- and post-compression and determine whether the tumor is benign or not with several statistical materials. The current results show that these proposed studies have great achievements. At last, we also propose several topics in future researches.
|
|---|
