Normal ultrasound values for pole-to-pole kidney length (LPP) are well established for children [1, 2], adults [3, 4], and seniors [5]. A short LPP usually allows chronic kidney failure to be easily distinguished from acute kidney failure with normal or enlarged values [6, 7]. Moreover, changes in LPP, parenchymal width (PW), cortex width (CW), or volume can be associated with atherosclerotic renal disease [8], arterial hypertension [9], atherosclerotic renovascular disease [10], or diabetes mellitus [11], or be indicative of these. The renal dimensions also allow conclusions as to the single kidney glomerular filtration rate to be made [12]. Aside from the acceptable reliable estimation of the LPP by ultrasound [9], all dimensions can now be easily determined by MRI [10], CT [8], and probably ultrasound as well. However, thus far no normal values exist, and the few known LPP values of healthy persons [8, 10] are different from the ones indicated by ultrasound [4]. The objectives of this study were thus to establish normal CT values for kidney dimensions from a group of unselected patients and identify potential influencing factors.

Although ultrasound is the primary method of choice for examining the kidneys, we preferred spiral CT for this study because on the one hand, some parameters were not measurable by ultrasound, and on the other hand, individual reformatting in all dimensions with resolutions up to the sub-millimeter range could be made from the volume data sets retrospectively, allowing to anticipate exceedingly precise measurements. Depending on the contrast medium phase, it is possible to delineate the renal structures and collecting system exactly and assess the surrounding organs precisely. Some of the influencing factors indicated for the LPP were already known, others, such as the number of vessels or position of the kidney were not, although their effect is sometimes pronounced and certainly should be taken into account when assessing size in individual cases.

The values for LPP correspond very closely with those indicated for ultrasound [4]. In addition to the assessment of coronal reformatting, the most reliable method of determining the LPP [14], sagittal and individual reformatting were also used, depending on the rotation of the kidneys, so that we must assume the greatest possible reliability of the results. The thin slice, multidetector technique has been chosen in order to obtain very accurate data.

Yet we must point out a few weaknesses of the study.

The patient group was not a randomly selected sample. The alternative, selecting a real random sample, was not possible due to the danger from radiation exposure from the CT. The extent of a potential bias effect, if any exists, should be minimized by the large number of patients with various illnesses unrelated to the kidneys and lower urinary tract. In order to estimate the extent of their influence, diseases of the kidney and lower urinary tract apparent in the imaging were not excluded if they were not known or symptomatic in the patient history and did not affect the renal parenchyma. Omitting the data from these patients did not lead to a change in the kidney sizes in the area indicated. Due to the exclusion of clinically conspicuous "maximum variants" of these diseases we can assume only that the influences subordinate to the main factors were somewhat underestimated. Patients in the initial stages of chronic kidney disease might have been included in the patient group accidentally; however, they were probably under-represented due to negative selection by the choice of the scanner in comparison to a completely randomly selected sample. After a diagnosis was made, each CT was evaluated only once during the study, and not, as would have been desirable, twice by two different observers. To minimize errors, this evaluation was carried out by two observers in consensus. The important issue of transferability of the values to ultrasound cannot be addressed for most of the measurements, but the almost exact correspondence of the LPP values with ultrasound data [4] can be considered a good indication of transferability.

The main advantage of a MD-CT over a single-row spiral CT is a shorter acquisition time. Thus, the volume data set from a MD-CT is much less subject to motion or breathing artifacts. It is known from experience, that data from single-row spiral-CT are rather accurate as well, but with a view to the fact that comparative studies do not exist, and cannot be performed due to ethical considerations, one can only hypothesise that the findings are repeatable using single row technique. Volumetric analysis was not used, as it is very time-consuming and expensive [6] and its application has not yet become established. Furthermore, we did not use the body surface area (BSA), a well known influencing factor for kidney volume, [15–17], as the more common BMI was used as a criterion for the factor "obesity" [18], and the two indexes are not independent of one another. Moreover, apart from the fact, that the creatinine levels were normal in most of the patients, or < 2 mg/dl in a small subgroup of patients, respectively, nothing is known about possible correlations of kidney function and size measurements in the present study. Results from ultrasonographic volume measurements of kidneys are promising [19], but further research is necessary in order to improve these opportunities.

In principle, kidney length can be estimated using ultrasound, MRI, intravenous pyelograms and CT, amongst others [14]. The CT predicts kidney length better than other modalities, but all modalities are connected with prediction errors in view of the kidney length [14]. The existent CT data regarding the kidney length were anticipated to be improved considerably, because they stemmed from 7 mm thick slices [14], in the worst case bringing about an estimation error of as much as 14 mm in the z-axis due to partial volume artefacts. However, the consistency of the data from our study, compared to the study of Kang et al. [13] has to be regarded as good, and even the values of the standard deviation are in agreement. The values presented here are for kidney length are slightly higher, because the longitudinal axis was exactly adjusted in every single kidney in 3D, thus correcting the underestimation of length due to projection errors in the x and y axes. This underestimation of length is the main problem when using ultrasound as well, as can be inferred from the data of Kang et al. [14]. The longitudinal axis of a kidney is not always perfectly adjustable in ultrasound. Moreover, the ultrasound technique is dependent on the sonographer [20]. Only little is known about kidney size measurements using MRI. It seems to be better than ultrasound in terms of estimating the kidney length [21]. Though a very high agreement with data obtained from CT may be expected, a study comparing CT and MRI in this regard is currently missing.

The differentiated observation of kidney sizes is of great significance clinically, as many diseases are associated with changes in kidney size [4]. The normal range is large [22], and what is "normal" depends on many factors. Within the standard deviation of LPPs there are values <9 cm in slim older women and up to 13 cm for men in their fifties. In the presence of other factors such as normal ADRAs or conspicuous body size, there are cases where LPPs of <8 or >14 cm can be considered normal and not be misunderstood as a sign of a cirrhotic kidney or acute kidney failure. Non-gender-specific data, according to which normal right kidneys are 11 ± 1 cm and left kidneys 11.5 ± 1 cm long [23], or 11 and 12 cm long, 5 and 7 or 7.5 cm wide, and 2.5 or 3 cm thick [24, 25], are not particularly useful in clinical practice. The influencing factors for size must be viewed individually to arrive at any relevant conclusions and information.

The age-related decrease of the LPP [4] and PW [26] are well known. The increase of the LPP in men up to their fifties has already been documented in data by Simon [27], who, however, did not consciously record it. While the LPP in men is only slightly larger than for women in the third decade, after the fifth decade it is about 10 mm longer, i.e. 10% of the LPP. We assume that sex hormones influence this. Be that as it may, with respect to the nonlinear relation of LPP to age in men, we waived the comparative gender-specific estimation of the influence of age based on various linear and nonlinear models, as it would have been of dubious value. It should be noted only that age is the greatest negative influencing factor on both CW and PW.

The great influence of BMI on LPP, CW, and PW was anticipated due to the known influence of body weight and BSA [6]. It appears to be more pronounced in women. The differences in the average values of clinically and morbidly obese patients in comparison with patients of normal weight [28–30] were as much as 20%.

The influence of height on LPP is also well documented [4] - it is by far the greatest independent predictor. Its influence on the PW is approximately comparable to that of BMI, but strangely enough, there is no influence on the CW. This could facilitate arriving at conclusions as to the extent of arteriosclerotic renal disease based on the CW [8].

The factor "stenosis of the renal arteries" is also well known [8]. The part it plays in the models with respect to the LPP is somewhat less pronounced than height, BMI, or gender. This is compatible with the observation that within a natural course of 33 months, only 16.2% of kidneys in patients with at least one renal artery stenosis are reduced in length by at least 1 cm - after exclusion of 7.7% of the kidneys in the patient group with either renal artery occlusion or atrophy [31]. Thus, less than 8% of the kidneys with arterial stenosis are found to be atrophic. Due probably to a partially collinear effect with age with respect to CW and PW, this factor ceased to be an independent predictor on one side of the models. Furthermore, the known independent and also age-related atrophogenic effect of arterial hypertension may also play a part [31].

The influence of the position of the kidneys on length is conspicuous - the further cranial and dorsal a kidney is, the longer it is. The former is known for the special case of pelvic kidneys [32]. We have no explanation for this strong effect, which surpasses even that of a renal artery stenosis. Since the length of the shortest perpendicular from the dorsal kidney surface to the dorsal fascia, as an indicator of perirenal fat, correlates positively with the thickness of the renal capsule and the BMI, the opposite was expected.

Duplex kidneys are longer than the contralateral kidneys [33]. They are supplied by ADRAs more frequently than "normal" kidneys [34]. The possible more likely persistence of vessels in relation to the length of the kidney is a plausible, although hypothetical explanation for the pronounced independent positive interrelation between LPP and the number of vessels.

Since kidneys do not increase in width in women, and in men get only slightly wider with increasing age, while the renal pelvis widens greatly in both men and women, the RPRP becomes accordingly smaller. It can therefore be assumed that renal tissue is replaced by fat. This would make the RPRP ultimately a measure for the atrophy of the organs, contrary to the currently valid opinion [4]. Whether and under what circumstances cortical or medullar tissue is replaced must be the subject of future research.

In summary we can say that there is a wide range in kidney sizes. LPPs of less than 8 cm as well as more than 14 cm can be normal. Among the strong positive influencing factors on LPP, CW, and PW in adults, the BMI, male gender, and height are most important. Cranial and posterior position of the kidneys is also significant. ADRAs have a strong positive effect. With respect to the cortex, the absence of the contralateral kidney is by far the most important factor. The most important negative influencing factors are age and stenosis of the renal artery. Due to the complex interaction of the factors described, the sizes presented here should serve as a reference point for the difficult differentiation between what is normal and pathological in individual cases in clinical routine.