信噪比(成像)
荧光
背景噪声
计算机科学
医学
人工智能
核医学
生物医学工程
物理
光学
电信
作者
Jacqueline van den Bos,Rutger M. Schols,Sander M. J. van Kuijk,Fokko P. Wieringa,Laurents P. S. Stassen
出处
期刊:Journal of Laparoendoscopic & Advanced Surgical Techniques
[Mary Ann Liebert, Inc.]
日期:2019-07-01
卷期号:29 (12): 1549-1555
被引量:3
标识
DOI:10.1089/lap.2019.0103
摘要
Aims: To investigate whether different calculation methods to express fluorescence intensity (FI) as target-to-background (BG) ratio are comparable and which method(s) match with human perception. Materials and Methods: Comparison of three calculation methods from current literature (OsiriX®, ImageJ®, and Photoshop®) to objectify FI during laparoscopic cholecystectomy measured at the exact same locations within recorded images of two categories: ex vivo and in vivo. Currently applied formulas to present FI in relation to the BG signal are compared with the subjective assessment by the human observers. These three formulas are Signal contrast = (FI in fluorescence regions-FI in BG)/255; Target-to-background ratio = (FI of target-FI of BG)/FI of BG; Signal-to-background ratio = FI of cystic duct/FI of liver and Target-to-background ratio = (FI of target-noise)/(FI of BG-noise). Results: In our evaluation OsiriX and ImageJ provided similar results, whereas OsiriX values were structurally slightly lower compared with ImageJ. Values obtained through Photoshop were less evidently related to those obtained with OsiriX and ImageJ. The formula Target-to-background ratio = (FI of target-noise)/(FI of BG-noise) was less corresponding with human perception compared with the other used formulas. Conclusions: FI results based on measurements using the programs OsiriX and ImageJ are similar, allowing for comparison of results between these programs. Results using Photoshop differ significantly, making direct comparison impossible. This is an important finding when interpreting study results. We propose to report both target and BG FI in articles, so that proper interpretation between articles can be made.
科研通智能强力驱动
Strongly Powered by AbleSci AI