Model-Based, Noninvasive Monitoring of Intracranial Pressure
Abstract
The project aims were (i) to further develop and validate a novel model-based approach to noninvasive, calibration-free determination of intracranial pressure (ICP) from quantities that are routinely measured in clinical settings, and (ii) to initiate the creation of a publicly available and growing reference database of physiological signals collected from brain injury patients. A physiologically based model relates ICP to simultaneously measured waveforms of arterial blood pressure (ABP), obtained via radial artery catheter or finger cuff, and cerebral blood flow velocity (CBFV) at a major cerebral artery, measured by transcranial Doppler (TCD), thus enabling ICP to be estimated. Our initial target population comprised subarachnoid hemorrhage patients in the neuro-ICU at our partner hospital, for whom ICP, ABP and CBFV are currently measured as the clinical standard of care. The project s major accomplishments include: assembling a suitable system for data collection in the neuro-ICU; specifying a careful data collection protocol; arranging for the collected data to be transferred to the publicly accessible MIMIC II database; converting our earlier batch-mode ICP estimation algorithms to run in real time; and, on 28 windows of data processed thus far, obtaining ICP estimates with mean bias -0.7 mmHg, standard deviation of error 4 mmHg, and root mean square error 3.9 mmHg. These results are twice as good as in our earlier validation studies.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 2013
- Accession Number
- ADA602081
Entities
People
- George C. Verghese
Organizations
- Massachusetts Institute of Technology