Evaluation of automated microvascular flow analysis software AVA 4: a validation study

Abstract Background Real-time automated analysis of videos the microvasculature is an essential step in development research protocols and clinical algorithms that incorporate point-of-care microvascular analysis. In response to call for validation studies available software by European Society Intensive Care Medicine, building on a previous study sheep, we report first human AVA 4. Methods Two retrospective perioperative datasets microcirculation (P1 P2) one prospective healthy volunteer dataset (V1) were used this study. Video quality was assessed using modified Microcirculation Image Quality Selection (MIQS) score. Videos initially analyzed with (1) 3.2 two experienced investigators gold standard semi-automated method, followed (2) 4.1. Microvascular variables measured perfused vessel density (PVD), total (TVD), proportion vessels (PPV). Bland–Altman intraclass correlation coefficients (ICC) measure agreement between methods. Each method’s ability discriminate microcirculatory states before after induction general anesthesia paired t-tests. Results Fifty-two from P1, 128 P2 26 V1 met inclusion criteria Correlational revealed poor no 4.1 3.2. Following anesthesia, TVD PVD increased significantly P1 ( p < 0.05) 0.05). However, these changes could not be replicated data generated Conclusions suitable tool or purposes at time. Future flow should aim new software’s standard, its thresholds which performance becomes unacceptable.