http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
- 中文 을 입력하시려면 zhongwen을 입력하시고 space를누르시면됩니다.
- 北京 을 입력하시려면 beijing을 입력하시고 space를 누르시면 됩니다.
RISS 인기검색어
- 검색키워드
- 저자 : Kaifi, Reham Essam (검색결과 1 건)
-
Kaifi, Reham Essam ProQuest Dissertations & Theses The George Washing 2020 해외박사(DDOD)
Femoral artery pseudoaneurysms (FAPs) are a major source of morbidity after cardiac catheterization. Color duplex ultrasound (US) is the gold standard to identify FAP neck and to confirm the diagnosis. Nevertheless, evaluation of anatomical features such as sac and neck size from B-mode US are equally important because follow-up observations play an important role in determining proper treatment. However, US is an operator-dependent modality, potentially resulting in high levels of intra- and interobserver variability. Therefore, it is important to know the range of variability for operators who provide follow-up US examinations to keep track of the actual changes in FAP measurements and to increase the chance for making the correct treatment decision.The aims of this dissertation were to perform image segmentation on B-mode US images to detect FAP; to quantify the intra- and interobserver reliability (the assessment of the consistency of measurements) and variability of FAP measurements between pre- and postprocessed US images acquired from two data sets; and to report the performance metrics (sensitivity, specificity, and Dice similarity coefficient) of the proposed image analysis by comparing the segmentation results of the proposed algorithm with the repeated estimates of “truth,” performed by expert sonographers.Retrospective image analysis based on histogram equalization, median filter, Otsu’s adaptive threshold, and region growing was conducted on 48 US images. Images were evaluated by four observers (two experts and two novices). Observers measured FAP sac and neck from pre- and postprocessed images; they then remeasured again after two weeks. The intra- and interobserver reliabilities were evaluated by calculating the intraclass correlation coefficient (ICC). The intra- and interobserver variability was calculated by using the mean absolute difference (MAD) between the first and the second measurements. Smaller MAD values point to less variability. For FAP neck length measurements, the average reliability (which is defined by ICC value) for expert observers from unprocessed images was 0.90, and the ICC increased to 0.97 from processed images, indicating an improvement of 7.7%. By contrast, the average reliability ICC value for novice observers taken from unprocessed images was 0.44, indicating poor reliability; this increased to 0.83 from processed images. indicating good reliability with an improvement of 88.6%. Paired t-tests for novices showed significant difference between trials when measurements were taken from unprocessed images with p-value < 0.05. In addition, the average measurements variability MAD value of FAP neck length taken by experts from unprocessed images was 0.18 cm, which then reduced to 0.12 cm from processed images. By contrast, the average measurements variability MAD value for novices from unprocessed images was 0.29 cm, which then reduced to 0.15 cm from processed images. The processed images lead to more reliable (consistent) measurements by the two novice observers in this study.In addition, the results of the performance metric (Dice overlap similarity coefficient) for region growing for the following cases: original B-mode, after histogram equalization, after the median filter, and after both enhancements were 0.87, 0.84, 0.87, and 0.90, respectively. The coefficient ranges from 0 to 1, where 1 represents perfect agreement and 0 represents no agreement. This indicates that applying region growing after both enhancements (histogram equalization followed by the median filter) showed better agreement when overlapped with manual segmentation compared to other cases if applied independently with a significant difference p-value < 0.05.
연관 검색어 추천
이 검색어로 많이 본 자료
활용도 높은 자료
