Objective interrogation of signal presentation from surgical near-infrared fluorescence systems for user and computerised interpretation

Clinical Biophotonics II(2022)

引用 1|浏览14
暂无评分
摘要
Near Infrared (NIR) camera systems in conjunction with Indocyanine Green (ICG) have been clinically applied for intraoperative tissue perfusion assessment. However signal interpretation remains human operator dependent, prone to inter-user variability and to distance-related signal attenuation. In this work, fluorescence signal emission from ICG-human albumin solutions were quantified at a variety of camera-target distances and speeds of camera movements. Utilising an electromagnetic field generator and sensors, these characteristics were spatially resolved and logged in real time. Such spatial mapping capability was also investigated in a clinical operation to confirm clinical applicability. Four clinical (two open, Elevision Medtronic, EMO, Ireland and Spy-Phi, Stryker, SSO, US and two laparoscopic, Elevision, Medtronic, EML, Ireland and Pinpoint, Stryker, PNL, USA, with both zero and 30 degree lenses) NIR systems were evaluated. Fluorescence imagery from each system regarding wells filled with ICG solutions were compared using bespoke region-of-interest (ROI) annotation, tracking and quantificative software with the cameras of each both still and moving at different distances from these targets. Chronologically-synchronised logged data featuring spatial coordinates and fluorescence intensities were plotted. Subsequently distance-intensity scatters were synthesised on which appropriate curves were fitted. Peak intensity, ideal optical (field of view) distance (IOD), and peripheral image fluorescence signal loss were defined. In vivo, tumour-to-scope distance spatial mapping capability was demonstrated with a single sensor setup during transanal fluorescent endoscopic examination of a rectal tumour. Three systems (SSO, PN, EML) displayed inverse distance intensity curves which were fitted to inverse square functions while EMO demonstrated static intensity at all distances. EML displayed directionality with a sigmoid curve when moving towards the wells and an inverse square function moving away. The IOD for the thirty-degree scope was further away (PNL 24.7cm vs 14cm) than the 0 degree despite having similar distances where their intensity maxed out (10cm and 8.27cm for the 30 and 0 degree respectively). While all laparoscopic systems displayed peripheral vs central well fluorescence signal hypoattenuation (p<0.001), open systems (EMO and SSO) were shown to hypoattenuate intensity centrally versus peripherally. In vivo assessment successfully logged a tumour to camera distance of 5.3cm during real-world clinical use. Fluorescence detection performance differs between systems in situations corresponding to real-world use. Such behaviours can be mathematically described and modelled enabling system-specific best use guidance and also should inform algorithms developed for the purpose of automated quantitative/artificially intelligent ICG perfusion angiogram classification.
更多
查看译文
关键词
Indocyanine Green, Machine learning, ICG, fluorescence angiography, near infra-red
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要