158 Chapter 8 Data quality assessment Sequences were inspected for artefacts, such as erroneous ROI tracking or surgical instruments blocking a clear view on the intestinal tissue. Artefacts were excluded from further analysis. The middle 96 frames of each recording were used to equalize sequence length and prevent any selection bias. ROIs were placed, based on surgical pen markings, and measured 60x60 pixels. Cut-off values In addition to real-time quantitative perfusion values, a cut-off value for LSPUs was calculated, aiding in the identification of well-perfused and ischemic tissue. To estimate this cutoff value, the Youden index was used; a measure for evaluating the effectiveness of diagnostic tests based on sensitivity and specificity. The index ranges from 0 to 1, with a value of 1 indicating a perfect test with no false positives or false negatives, and a value of 0 indicating a test that performs no better than chance 26. First, a cut-off value with the highest Youden Index for LSPU was calculated, based on anatomically ischemic and well-perfused tissue. Tissue with LSPU levels below the cut-off value were classified as ischemic. The same process was repeated to calculate the cut-off value for lactate levels, which was compared to existing evidence to validate the LSPU cut-off calculation. Inter-observer reliability In this study, the selection of ROIs was done subjectively by the surgeon based on real-time interpretation of the color map on the monitor. To assess the robustness and reliability of color map interpretation, an inter-observer variability analysis was performed after the experiment using the LSCI images taken during the surgery. Five LSCI experts and five physicians (surgical residents) with no experience in assessing LSCI images were asked to locate the watershed areas on the LSCI colourmaps, as well as the ischemic and wellperfused areas. The inexperienced physicians received an introductory training consisting of two slides on how LSCI works and how to interpret the colormaps, along with one training image and accompanying text on how to place ROIs. Inter-observer distance was registered in centimetres between all assessors, measuring over a midline on the small intestinal loop. A more detailed explanation can be found in supplemental 1. An expert was defined as someone having multiple years of experience in working with or developing LSCI systems. Statistical analysis Microsoft Excel (Microsoft Excel version 2312, Microsoft Corporation, Redmond, Washington, United States) and IBM SPSS statistics software package (IBM SPSS statistics version 27, IBM Corp, Armonk, New York, United States) were used to perform statistical analyses. A linear mixed effect model was built using a random intercept model with time and ROI location as fixed effects and an interaction term between time and ROI. A scaled identity covariance structure was used, and it was considered adequate to use Restricted Maximum Likelihood (REML) as estimation method, to prevent biasing by the used method. Logarithmic curve
RkJQdWJsaXNoZXIy MTk4NDMw