Diffuse Optical Spectroscopy Aided Non-Invasive Intracranial Pressure Estimation and Cerebral Autoregulation Assessment

In many common brain diseases, including traumatic brain injury, stroke, and hydrocephalus, intracranial pressure (ICP) can rise and lead to malperfusion of brain cells and ischemia. Furthermore, the brain’s ability to regulate cerebral blood flow despite changes in cerebral perfusion pressure (CPP) can be impaired. Measuring and controlling ICP to maintain a stable oxygen supply to the brain is therefore of high clinical importance. Current devices for ICP measurements are highly invasive and require lumbar puncture or the placement of a catheter into the brain through craniotomy. This thesis offers alternative methods to measure ICP using non-invasive, diffuse optical devices to gain information about cerebral blood flow and hemoglobin concentration changes in the brain. We propose to use a non-parametric transfer function approach applied to oxygenated hemoglobin concentration changes in the brain

to estimate fluctuations of ICP. Additionally, the cardiac pulse shape of cerebral blood flow was found to be associated with quantitative ICP. A machine learning approach is proposed that uses descriptive morphological features of the cardiac pulse to estimate the underlying ICP. We were able to show that the proposed methods perform

well in non-human primates under controlled manipulations of ICP and arterial blood pressure. The non-invasive ICP assessment further allows for cerebral autoregulation assessment, which otherwise often requires invasive ICP recordings. In order to demonstrate this, we performed a non-human primate study where we were able to show

that impaired cerebral autoregulation is largely driven by CPP, which is a function of blood pressure and ICP, further highlighting the need for non-invasive ICP measurements.

The discussed methods have the potential to create a lasting impact on ICP acquisition, not only for intensive clinical care, but also for currently inaccessible research on healthy volunteers. Lastly, a clinical translation to the pediatric intensive care is discussed and preliminary results of translation to human subjects are presented. The methods developed in this thesis have the potential to eliminate the need of invasive ICP sensors and therefore may help clinical decision making for treatment guidance in a variety of diseases.