Determination of total body water by multifrequency bio-electric impedance: development of several models.

B.K. van Kreel*, P.L. Cox-Reijven, P.B. Soeters

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review


Department of Clinical Chemistry, University Hospital Maastricht, The Netherlands.

Multifrequency bio-electronic impedance analysis (MF BIA) measurements are taken from a heterogeneous group of patients, varying in size between obese and slim. The measuring system uses four electrodes: two current and two potential electrodes. Three new models are developed to calculate total body water (TBW) from the BIA data, and the resulting TBW values are compared with TBW determined by D2O dilution. The results demonstrate that the most simple model provides the best TBW values. For individual patients, TBW can be determined by means of bioimpedance measurement with an accuracy of 3 litres. In the most simple model (model 1), the body is electrically represented by a cylinder, and corrections are made for the amount of fat. This is an extension of the model used by Xitron. In the more advanced models (2 and 3), the body is represented by a cylinder for the trunk, and truncated cones represent the arms and legs. In model 2, delta TBW amounts to 3 litres. It is shown that the resistance of the trunk is proportional to the square root of the length. In model 3, it is assumed that subcutaneous fat is a poor conductor of electric current. An equation is developed that describes the partition of subcutaneous fat, and the fat layer is then removed from the cones representing arms and legs and from the cylinder that models the trunk.
Original languageEnglish
Pages (from-to)337-345
Number of pages9
JournalMedical & Biological Engineering & Computing
Issue number3
Publication statusPublished - 1 Jan 1998


Dive into the research topics of 'Determination of total body water by multifrequency bio-electric impedance: development of several models.'. Together they form a unique fingerprint.

Cite this