Noninvasive assessment of intracranial pressure using functional matrix estimation method

Abstract

Intracranial pressure (ICP) reflects the neurological condition in patients suffering from a traumatic brain injury (TBI). Though considered as one of the most important parameters in neurocritical care, the acquisition of ICP involves invasiveness, with a considerable risk of complications. The presented study devised a novel method of acquiring ICP non-invasively via functional matrices, which usess arterial blood pressure (ABP) and cerebral blood flow velocity (CBFv) as inputs. The continuous ICP recordings of 193 TBI patients were subjected to the analysis. The relationship between ABP, CBFv and ICP was encoded within functional matrices during the five minutes of a learning period. Estimated ICP recordings were compared to the original, invasively measured ICP. The results showed that the single optimized function matrix method yielded a mean of absolute error (MAE) of 3.13 mmHg, with the SD of error (SDE) of 2.94 mmHg and the root mean square error (RMSE) of 4.40 mmHg. The multiple selective functional matrices method yielded MAE of 2.32 mmHg, SDE of 1.76 mmHg, and RMSE of 2.93 mmHg. The accuracy and low computational complexity of the methods employed in this study deserves attention.