Analysis of brain and cerebrospinal fluid volumes with MR imaging. Part I. Methods, reliability, and validation

Radiology. 1991 Jan;178(1):115-22. doi: 10.1148/radiology.178.1.1984289.


A computerized system was developed to process standard spin-echo magnetic resonance (MR) imaging data for estimation of brain parenchyma and cerebrospinal fluid (CSF) volumes. In phantom experiments, the estimated volumes corresponded closely to the true volumes (r = .998), with a mean error less than 1.0 cm3 (for phantom volumes ranging from 5 to 35 cm3), with excellent intra- and interobserver reliability. In a clinical validation study with actual brain images of 10 human subjects, the average coefficient of variation between observers for the measurement of absolute brain and CSF volumes was 1.2% and 6.4%, respectively. The intraclass correlations for three expert operators is greater than .99 in the measurement of brain and ventricular volumes and greater than .94 for total CSF volume. Therefore, the authors believe that their technique to analyze MR images of the brain performed with acceptable levels of accuracy and reliability and that it can be used to measure brain and CSF volumes for clinical research. This technique could be helpful in the correlation of neuroanatomic measurements to behavioral and physiologic parameters in neuropsychiatric disorders.

Publication types

  • Research Support, Non-U.S. Gov't
  • Research Support, U.S. Gov't, P.H.S.

MeSH terms

  • Algorithms*
  • Brain / anatomy & histology*
  • Cerebrospinal Fluid*
  • Humans
  • Image Processing, Computer-Assisted*
  • Magnetic Resonance Imaging / methods*
  • Models, Structural
  • Observer Variation
  • Software Validation