Variability Reduction in NIRS Measurements for Reliable TBI Patient Monitoring

Traumatic brain injury (TBI) is a leading cause of mortality and morbidity, commonly causing intracranial hematomas as a serious complication. While computerized tomography (CT) scans are the standard diagnostic tool for hematoma detection, their use for continuous monitoring is limited due to radiation risks and logistical difficulties. Near-infrared spectroscopy (NIRS) offers a noninvasive, portable, and cost-effective alternative, as evidenced by the Infrascanner—the first handheld device proven in clinical studies for hematoma detection. Despite its success in hematoma detection, the Infrascanner’s efficacy in tracking hematoma expansion remains unexplored. Our recent pilot studies with repeated recordings conducted to investigate Infrascanner’s potential as a monitoring device showed large signal variability (SD) even under unchanging or healthy conditions when its original protocols developed for hematoma detection are used. In this work, our primary goal was to develop novel data processing algorithms with a modified data collection regime and evaluate their performance in reducing SD in comparison to each other and the original algorithms. Our results suggested that among the proposed algorithms, the most promising was based on eliminating maximal values from hemispheric measurement pairs, resulting in significantly improved signal stability and overall device performance. Our key contributions in this study involve the development and evaluation of novel data collection and processing methods for the Infrascanner to reduce its SD and false alarm rate (FAR) while improving its sensitivity and correlation with hematoma size, which can further strengthen the potential utility of the Infrascanner in monitoring TBI patients for hematoma expansion.