History
Early workers in the field of clinical nutrition used standard x-ray techniques to examine fat layer thickness. The high radiation exposure and low image contrast limited the research and clinical applicability of this approach. In 1973, Geoffrey Hounsfield and his colleagues introduced the first computerized axial tomography (CT) system, and by the early 1980's, CT scanners were installed in hospitals throughout the world. The early CT approach provided cross-sectional images with high contrast and by 1979, the first reports of skeletal muscle measurement appeared. The first measurements of visceral organ volumes were reported in 1979, as well as estimates of visceral adipose tissue in 1981. Several contiguous image slices were assembled into the complete three-dimensional adipose tissue or organ compartment of interest. While CT was a major breakthrough in quantifying the volumes of tissues and organs, applicability was limited by radiation exposure. Within a decade, the first reports of magnetic resonance imaging (MRI) of humans appeared and Foster et al. in 1980s’reported the first MRI body composition studies. As with CT, MRI provides the unique capability of quantifying tissue and organ volumes in vivo but without radiation hazard. Gradual advances in both CT and MRI capabilities now make single or multiple slice tissue, as well as organ analysis, a reference approach against which other techniques can be compared to.

Application
Both CT and MRI provide high-resolution cross-sectional images through selected anatomic regions. At one extreme the entire body can be imaged and the volume of all major tissue-system level components estimated. Volume estimates can be converted to mass values by assuming specific tissue densities. Depending on the selected slice number, whole-body evaluations require scan times ranging from about 20 minutes to several hours. As CT exposes subjects to radiation, there are several reports of whole-body CT studies in humans. Moreover, CT and MRI in phantom and cadaver studies provide similar tissue volume estimates. The trend today is to apply MRI whenever possible. Specific aspects of scanning protocols are reported in earlier studies.

Once images are collected, analysis can take one of several different pathways. For CT, pixel intensities are designated in Hounsfield units (HU), and calibrations are similar among all CT scanners. Hounsfield unit ranges differ between tissues, notably adipose tissue, lean soft tissues, and bone vary sufficiently in pixel intensity to allow component separation using designated HU ranges. Hounsfield unit ranges for adipose tissue, muscle/organs, and bone are: -190 to –30 HU; -30 to +100 HU; >100 HU.

Selected organs and tissues can also be traced directly on the CT scanner console and related areas within each slice established. As CT imaging time is usually rapid (several seconds per slice), images of moving objects such as the viscera, secondary to peristalsis, and respiration, are still relatively sharp and boundaries are clear.

Analysis of MRI scans is more complex as pixel intensity varies according to the selected imaging sequence and other factors. Standard pixel “ranges” cannot therefore be set, and analysis is on a scan-by-scan basis. Dedicated image analysis software is usually applied rather than standard system radiology software. Image acquisition times for MRI are usually longer than they are for CT, and patients should maximize their breath holds and maintain a stable position during the scan. Images of the viscera tend to be less sharp than they are for CT, although new MRI scanning sequences are improving image clarity. Gated MRI scans allow development of high contrast images of the myocardium. For both CT and MRI, training is required for image analysis and reading times can vary from several minutes for a single slice to several days for a whole body.

Imaging methods, both CT and MRI, are uniquely capable of acquiring tissue-organ level volume estimates including all major organs and tissues, visceral adipose tissue, and regional estimates. CT and MRI estimates of visceral adipose tissue, either a single slice or multiple slices, are considered the reference against which other techniques are compared. Cost, instrument access, and the need for trained image analysis technicians may limit routine imaging method use to specialized research studies and centers. CT and MRI are not appropriate for use in field studies of body composition, although both methods can be used to “calibrate” or validate other simpler less costly methods.

While today we still focus mainly on produced images, rapid growth in magnetic resonance spectroscopy and functional MRI offer great promise in the study of human physiology and metabolism. It is likely that these new developments will not only add to our ability to quantify body composition, but to enhance of knowledge of closely-related metabolic processes as well