City Research Online

Abstract

Traumatic brain injury (TBI) is a leading cause of death and disability, often resulting in increased intracranial pressure (ICP) and cerebral ischemia. Current ICP measurement methods involve invasive, non-therapeutic procedures. This research aims to develop a non-invasive, continuous optical system for monitoring ICP and cerebral oxygenation. Using backscattered brain optical signals, it leverages cerebral pulsatile photoplethysmograms (PPGs) and non-pulsatile near-infrared spectroscopy (NIRS) signals to assess ICP and oxygenation. The innovation lies in using cerebral NIRS-PPGs to measure ICP, based on the hypothesis that changes in ICP affect cerebral PPG signal morphology. These changes in morphological features, with the support of advanced algorithms including Machine Learning (ML) models, could be utilised in translating the changes in the pulsatile signals in absolute measurements of ICP. The research firstly implemented Monte Carlo simulations to fully understand the effect of multi-source detector separations on brain light tissue interaction. Secondly, a novel reflectance, custom-made TBI multiwavelength and multisource-detector optical sensor and instrumentation, including advanced signal processing algorithms, was designed to acquire, pre-process, and analyse raw PPG signals (AC + DC) from the brain. Thirdly, a novel head phantom and an in vitro brain haemodynamic system were developed for evaluating the sensor. The phantom was the ideal tool for simulating different clinical scenarios that cannot be implemented in real in vivo studies. Fourthly, this research carried out three in vitro studies to investigate the sensor's capability to non-invasively monitor intracranial pressure and oxygenation. The first study evaluated the quality of the optical signals acquired from the developed probe at different source-detector (S-D) separations and multiple wavelengths. It was concluded that the optimal S-D separation to reach the cerebral tissue, and acquire good quality PPG signals, was within 3 cm and 4 cm. The second study assessed the central hypothesis of this research by recording PPG signals from the phantom’s brain at different intracranial pressure levels and implementing ML models utilising pertinent features from the PPG. Results from the second study showed a correlation coefficient of 0.86, mean absolute error of 3.7 mmHg, and limits of agreement of ±4 mmHg, which suggest that NIRS-PPG signals could estimate ICP non invasively. Finally, a third study demonstrated the sensor’s response to in vitro changes in blood oxygenation levels, with less than 33.8% error in half the measurements compared to the reference. This final implementation of spatially resolved spectroscopy algorithms actualize the proposed non-invasive multimodal monitoring sensor for traumatic brain injury. The novel technological developments and the new knowledge acquired from this research paves the way for the development of a transformative non-invasive optical sensor technology for the continuous monitoring of ICP and cerebral oxygenation in TBI patients and beyond.

Publication Type:	Thesis (Doctoral)
Subjects:	Q Science > QH Natural history > QH301 Biology T Technology > TA Engineering (General). Civil engineering (General)

Export

Downloads