Ruesch et al. demonstrate a machine learning technique to accurately estimate intracranial pressure (ICP) in macaque monkeys, utilizing non-invasive optical monitoring. ICP is a critical parameter in guiding intervention in the neuro-intensive care unit: ICP elevation can be fatal. In current clinical practice, cerebrospinal fluid drainage is often initiated when ICP > 20 mmHg, with attendant risks and morbidity. Current clinical ICP monitoring requires invasive probes which penetrate the brain tissue and are thus reserved for the most critically ill patients and deployed at a single site. Unfortunately, current non-invasive ICP monitors have limited accuracy and precision.

In this study, the authors utilized diffuse correlation spectroscopic measurements of the cerebral blood flow waveform in the microvasculature. Features of this pulsatile wave and the mean arterial blood pressure were used to train a regression forest in estimating ICP, which was experimentally manipulated by an external reservoir connected to a cerebral ventricle. These estimates correlated strongly with invasively measured ICP and had high sensitivity and specificity for ICP > 20 mmHg. Further developments of these diffuse optical devices have the potential to offer a reliable noninvasive ICP measurement, which would permit continuous, long-term monitoring, in multiple regions of the brain, and in a wide range of patients. Together, these advantages suggest that diffuse correlation spectroscopy could become the method of choice for monitoring ICP.

---

---

 Add Comment

You must log in to add comments.