Computed Tomography (CT)

There are a number of imaging techniques for determining total and regional body composition. Computed tomography (CT) and magnetic resonance imaging (MRI) provide cross-sectional images that can be used to determine total adiposity and are one of the most accurate tools available for directly quantifying body composition at a tissue level. As such, CT and MRI (discussed further in Magnetic Resonance Imaging) are often considered the criterion measures for assessing visceral fat and skeletal muscle in vivo. Other imaging techniques—such as dual energy x-ray absorptiometry (DEXA) and ultrasound—are also useful clinical imaging techniques for assessing total and visceral adiposity. The strengths and weaknesses of DEXA and ultrasonography are addressed in another section (see Other Imaging Techniques).

CT uses ionizing radiation and differences in tissue x-ray attenuation to produce cross-sectional images of the body [1, 2]. X-ray attenuation depends mainly on matter density and is commonly expressed in Hounsfield units (HU). Lower density tissues such as fat have lower HU ratings than higher density tissues such as muscle or bone. Cross-sectional CT images are composed of many pixels, each with a HU value reflecting the molecular composition of the tissue. Although CT is easier to use than MRI, radiation exposure makes it unsuitable for multiple-image whole body tissue quantification (Figure 1).

CT is one of the gold standard techniques used for in vivo quantification of skeletal muscle mass (Figure 2). Muscle mass and changes to it are related to muscle strength [9-11], and accurately determining skeletal muscle mass is particularly important in elderly populations, who are at increased risk of sarcopenia and functional impairment due to low muscle mass. Measures of skeletal muscle by a single CT image have been validated using cadaver measures and show a high level of agreement (R²=0.94, standard error of estimate=9.5%), with a coefficient of variation of approximately 2% [12]. When compared to cadaver values, CT error improved to approximately 1% when volume measures were acquired from multiple images. However, as CT involves radiation exposure, a single image at the mid-thigh is commonly used as a proxy measure of whole body skeletal muscle in both men and women (R²=0.77 to 0.79) [13].

MRI and CT are the only in vivo methods available to directly and accurately quantify visceral fat. Visceral fat is the fat located within the abdominal muscle wall that surrounds the organs (or viscera). On average, it accounts for only 12% and 5% of total body fat in men and women, respectively (Figure 3). As with skeletal muscle, measuring visceral fat with multiple images is costly, labour intensive, and in the case of CT, involves substantial radiation exposure. Consequently, visceral fat is normally assessed using a single MRI or CT image at L4-L5. However, it is important to note that visceral fat values obtained through CT are not necessarily comparable to those obtained through MRI [14,15].

There is a growing literature on the importance of visceral fat as a strong predictor of dyslipidemia [16-22], glucose tolerance [23,24], insulin resistance [24], and incidence of hypertension [18-20,25], cardiovascular disease [26], and type 2 diabetes [27]. Even within a given BMI category, men and women with greater amounts of visceral fat are more likely to be at increased risk of developing the metabolic syndrome [28], disturbances in glucose tolerance [29,30], and insulin resistance [29,31]. Further, prospective studies have indicated that visceral fat predicts future hypertension independent of age, BMI, weekly energy expenditure, systolic blood pressure, and glucose tolerance [25]. Similarly, visceral fat is a significant predictor of future type 2 diabetes independent of age, BMI, glucose intolerance, and family history [27]. It is not surprising, then, that visceral fat has also been linked to increased risk for all-cause mortality [32]. Accordingly, the International Atherosclerosis Society and the International Chair on Cardiometabolic Risk Working Group on Visceral Obesity has published a joint position statement emphasizing the importance of visceral adiposity and of ectopic fat depots and the risk of type 2 diabetes and cardiovascular disease [33].

Although excess visceral fat is increasingly recognized as a health hazard, there is little consensus as to optimal cutoffs for identifying individuals with excess visceral fat and therefore at increased health risk. Suggested cutoffs range from 100 to 130 cm². Regardless of optimal cutoff values, an increase in visceral fat clearly worsens the metabolic profile, even among normal weight individuals [34]. Consequently, individuals should try to minimize visceral adiposity and avoid increases in visceral fat regardless of cutoff values. Moreover, because it is not possible to routinely measure visceral fat using MRI and CT, individuals should focus on routinely measuring waist circumference, which is the best surrogate measure currently available.

CT can also be used to assess ectopic fat in the muscle and liver [35-38]. As measured by CT, liver fat infiltration is calculated by determining the attenuation values for each voxel within a region of interest in the liver. CT attenuation values depend on the molecular composition of the tissues within each voxel. Fat has a lower density than water and protein, and liver fat infiltration is reflected by a lower liver density and thus lower attenuation values [36]. However, normal and fatty liver values overlap [39]. As such, it has been suggested that liver density (CTL) should be expressed relative to the spleen attenuation value (CTS), which is not infiltrated with fat (Figure 4) [39]. In normal individuals, liver and spleen attenuation values have a constant relationship. The liver, however, is a denser organ and therefore has a higher attenuation value. A liver-to-spleen attenuation (CTL/CTS) ratio of less than one therefore indicates fatty infiltration [36,39]. Normally, the liver and spleen mean attenuation values are based on two or three regions of interest within the liver and spleen. However, due to the small area of interest and subjectivity involved in determining the regions of interest, the whole liver and spleen surface areas should be used to determine respective mean attenuation values (Figure 4) [40]. Attenuation values within the liver and spleen are fairly homogeneous throughout, and using the whole surface area can slightly reduce the inter-observer coefficient of variation from 5.1% [41] to 2.9% [40]. However, it is difficult to obtain a CT image that contains both liver and spleen. Not only does the spleen’s vertical position vary relative to the liver, both organs also vary in terms of their position within the abdominal cavity. As a multi-image approach is not feasible because of excess radiation exposure [42], a single axial image at the T12-L1 inter-vertebral space may be the best landmark for assessing both liver and spleen attenuation, given that liver and spleen can be identified at that level in approximately 90% of the men and women studied [40].

The average HU or mean attenuation value of adipose tissue-free skeletal muscle voxels can also be used as an index of skeletal muscle lipid content (Figure 5) [38]. As with the liver, the lower the skeletal muscle mean attenuation value or the greater the number of low-density skeletal muscle voxels (e.g., 0-30 HU), the higher the skeletal muscle lipid content. However, unlike the liver, fat is stored both inside and outside the muscle cell. As such, CT muscle attenuation values reflect both intra-myocellular (IMCL) and extra-myocellular (EMCL) lipid content. Although similar, they are not analogous to intra-myocellular lipid values obtained through skeletal muscle biopsy or proton magnetic resonance spectroscopy.

In summary, CT is one of the criterion methods for measuring visceral fat and skeletal muscle mass. It can also be used to assess lipid infiltration in tissues such as muscle and the liver. However, CT is expensive and involves radiation exposure, which limits the routine use of this tool for assessing body composition and predicting obesity-related health risk in clinical practice.