Modeling Cerebral Blood Flow and Flow Heterogeneity From Magnetic Resonance Residue Data

• Published in: Journal of cerebral blood flow and metabolism Vol. 19; no. 6; pp. 690 - 699
• Main Authors: Østergaard, Leif, Chesler, David A., Weisskoff, Robert M., Sorensen, A. Gregory, Rosen, Bruce R.
• Format: Journal Article
• Language: English
• Published: London, England SAGE Publications 01-06-1999; Lippincott Williams & Wilkins
• Subjects: Adult; Algorithms; Arterioles - physiology; Basal Ganglia - blood supply; Biological and medical sciences; Cerebral Arteries - physiology; Cerebrovascular Circulation - physiology; Humans; Investigative techniques, diagnostic techniques (general aspects); Magnetic Resonance Imaging; Medical sciences; Models, Neurological; Nervous system; Radiodiagnosis. Nmr imagery. Nmr spectrometry; Vascular Resistance; Flow heterogeneity; Vascular models; residue detection; Magnetic resonance imaging; Cerebral blood flow; Humans; Human; Regional blood flow; Central nervous system; Medical imagery; Exploration; Models; Hemodynamics; Technique; Nuclear magnetic resonance imaging; Brain (vertebrata)
• ISSN: 0271-678X; 1559-7016
• DOI: 10.1097/00004647-199906000-00013

Existing model-free approaches to determine cerebral blood flow by external residue detection show a marked dependence of flow estimates on tracer arrival delays and dispersion. In theory, this dependence can be circumvented by applying a specific model of vascular transport and tissue flow heterogeneity. The authors present a method to determine flow heterogeneity by magnetic resonance residue detection of a plasma marker. Probability density functions of relative flows measured in six healthy volunteers were similar among tissue types and volunteers, and were in qualitative agreement with literature measurements of capillary red blood cell and plasma velocities. Combining the measured flow distribution with a model of vascular transport yielded excellent model fits to experimental residue data. Fitted gray-to-white flow-rate ratios were in good agreement with PET literature values, as well as a model-free singular value decomposition (SVD) method in the same subjects. The vascular model was found somewhat sensitive to data noise, but showed far less dependence on vascular delay and dispersion than the model-free SVD approach.