The goal of this PhD project is to characterize and classify intracranial pressure (ICP) signals in near-to-normal subjects using both machine learning and pattern recognition strategies, which will need to be explored.
ICP monitoring is used in the diagnosis and treatment of a wide range of neurological and neurosurgical disorders as well as a part of multimodal monitoring in patients with severe acute brain disease.
There is a range of pressures where the brain functions properly, but as the pressure exceeds this range in either direction, function begins to progressively deteriorate. This is the simplest way of looking at ICP - as purely a number that needs to stay in its proper place. Going beyond that, pressure wave patterns of the ICP signal lasting seconds to minutes can also be analyzed.
From clinical experience and visual inspection of the ICP signal, there appears to exist varying patterns and sequences that are not yet formally classified, but potentially could be used for classification purposes. Identification is difficult since it requires experience in looking at these signals and the methods are still not 100 % reproducible between researchers classifying the signals. Until now, classifications have mainly been conducted manually.
This project aims to automatically classify the ICP signals, and, where relevant, combine these with other clinical data about the patients. It will also include work on the development of validation procedures for ICP measurements, in addition to the participation in clinical measurements of ICP.
At the end of this project we will hopefully have found answers to questions such as: can any repeatable patterns be observed in the signals before predefined macro-patterns occur? Are the waves or any patterns characterized by certain frequencies? Are these patterns occurring simultaneously to other clinical symptoms?
The project is funded by the Novo Nordisk Foundation and carried out in collaboration with the CSF Study Group from Rigshospitalet.