Bicen A Ozan, West Leanne L, Cesar Liliana, Inan Omer T
Inan Research Laboratory, School of Electrical and Computer EngineeringGeorgia Institute of TechnologyAtlantaGA30332USA.
Pediatric Technology CenterGeorgia Institute of TechnologyAtlantaGA30332USA.
IEEE J Transl Eng Health Med. 2018 Apr 3;6:4100207. doi: 10.1109/JTEHM.2018.2815539. eCollection 2018.
Intravenous (IV) therapy is prevalent in hospital settings, where fluids are typically delivered with an IV into a peripheral vein of the patient. IV infiltration is the inadvertent delivery of fluids into the extravascular space rather than into the vein (and requires urgent treatment to avoid scarring and severe tissue damage), for which medical staff currently needs to check patients periodically. In this paper, the performance of two non-invasive sensing modalities, electrical bioimpedance (EBI), and skin strain sensing, for the automatic detection of IV infiltration was investigated in an animal model. Infiltrations were physically simulated on the hind limb of anesthetized pigs, where the sensors for EBI and skin strain sensing were co-located. The obtained data were used to examine the ability to distinguish between infusion into the vein and an infiltration event using bioresistance and bioreactance (derived from EBI), as well as skin strain. Skin strain and bioresistance sensing could achieve detection rates greater than 0.9 for infiltration fluid volumes of 2 and 10 mL, respectively, for a given false positive, i.e., false alarm rate of 0.05. Furthermore, the fusion of multiple sensing modalities could achieve a detection rate of 0.97 with a false alarm rate of 0.096 for 5mL fluid volume of infiltration. EBI and skin strain sensing can enable non-invasive and real-time IV infiltration detection systems. Fusion of multiple sensing modalities can help to detect expanded range of leaking fluid volumes. The provided performance results and comparisons in this paper are an important step towards clinical translation of sensing technologies for detecting IV infiltration.
静脉输液疗法在医院环境中很普遍，通常通过静脉输液将液体输送到患者的外周静脉。静脉输液外渗是指液体意外进入血管外间隙而非静脉（需要紧急处理以避免瘢痕形成和严重组织损伤），目前医护人员需要定期检查患者。在本文中，研究了两种非侵入性传感方式，即电阻抗生物传感（EBI）和皮肤应变传感，用于在动物模型中自动检测静脉输液外渗。在麻醉猪的后肢上模拟外渗情况，将EBI和皮肤应变传感的传感器放置在同一位置。所获得的数据用于检验利用生物电阻和生物电抗（源自EBI）以及皮肤应变来区分静脉输液和外渗事件的能力。对于给定的误报率（即误警率为0.05），皮肤应变传感和生物电阻传感对于2毫升和10毫升外渗液体量的检测率分别可以大于0.9。此外，对于5毫升外渗液体量，多种传感方式融合可实现检测率为0.97，误报率为0.096。EBI和皮肤应变传感能够实现非侵入性实时静脉输液外渗检测系统。多种传感方式融合有助于检测更广泛范围的渗漏液体量。本文提供的性能结果及比较是传感技术用于检测静脉输液外渗临床转化的重要一步。
