google ads

Everything you need to know about Computed Tomography (CT) & CT Scanning

Dual Source Computed Tomography: A Novel Technique to Determine Stone Composition

Brian R. Matlaga, Satomi Kawamoto, Elliot Fishman

James Buchanan Brady Urological Institute, The Russell H. Morgan Department of Radiology and Radiological Science, The Johns Hopkins University School of Medicine, Baltimore, Maryland

Received 11 December 2007, Accepted 31 March 2008, Available online 10 July 2008


Dual source computed tomography (CT), a novel technology that employs two different x-ray sources, may provide an image resolution that has not been possible with conventional single source CT. We tested the ability of dual source CT imaging to distinguish calcium oxalate (CaOx) stones from calcium phosphate (CaP) stones, and both types of calciferous stones from uric acid (UA) stones.


CaOx, CaP, and UA stones were placed in a tissue phantom and imaged with dual source multidetector CT at both 80kV and 140 kV. Hounsfield units (HU) of each stone were recorded for the 140 kV and the 80 kV datasets by drawing regions of interest by circle, hand-drawing, and using a volume calculation method. The differences between, and the ratios of, the HU obtained by the two energy sources were compared among the stone groups.


Between CaOx and CaP stones, HU differences (80kV-140kV difference, hand-drawing technique: CaOx = 270.7; CaP = 369.5; UA = 11.45; P = 0.020 for CaOx vs. CaP; P < 0.001 for CaOx and CaP vs. UA) as well as ratios (140kV:80kV ratio, hand-drawing technique: CaOx = 1.44; CaP = 1.51; UA = 1.04; P = 0.001 for CaOx vs. CaP; P < 0.001 for CaOx and CaP vs. UA) were significantly different. There were no significant differences in stone size among the groups.


Dual source CT has the ability to distinguish not only between calciferous stones and uric acid stones, but also among stones composed of different calcium salts. Further studies are warranted to confirm these findings in a clinical setting.

Noncontrast helical computed tomography (CT) is the preferred method for evaluating the majority of patients suspected of harboring urinary calculi.1 The widespread utilization of CT for this task is due in great part to its ability to rapidly locate the stone in question, gauge its size, and inform the clinician as to associated complicating features such as hydronephrosis. A number of investigators have studied the ability of helical CT to predict stone composition.2, 3, 4, 5, 6, 7, 8, 9, 10 and 11 In general, CT has been reported to consistently distinguish stones composed of uric acid (UA) from stones composed primarily of calcium. However, the issue of whether CT imaging can distinguish among stones composed of different calcium salts, such as calcium oxalate (CaOx) stones from calcium phosphate (CaP) stones, is an area of open investigation, as published studies have yielded conflicting results.

A recent innovation in CT is the development of dual source CT imaging, a novel technology that utilizes two x-ray sources, rather than a single x-ray source as in conventional helical CT. In addition to improved temporal resolution, dual source CT scanners have the ability to simultaneously operate the two x-ray sources at different energy levels, which has the potential to differentiate materials on the basis of their unique energy-dependent profiles. It has been reported that dual source CT imaging may allow tissue differentiation that has not been previously possible with single source CT.12 We performed a study to test the ability of dual source CT imaging to distinguish stones composed of one calcium salt from stones composed of another calcium salt, as well as from UA stones.

Material and Methods

Human kidney stones were obtained from a stone analysis laboratory (Beck Analytical Services, Indianapolis, Indiana). The composition of the stones was confirmed by analysis with microscopic visual inspection, chemical reaction, and Fourier transform infrared microspectroscopy. The stones selected for this study were composed of CaOx monohydrate, CaP (hydroxyapatite), and UA.

The intact stones were hydrated in water and placed in a carotid anthropomorphic vascular phantom (Shelly Medical Imaging Technologies, Ontario, Canada). A phantom was employed to simulate in vivo conditions, as CT analysis of stones can be influenced by surrounding media. This phantom is a 5 × 21 × 2.5 cm structure, with an 8 mm diameter tubular lumen in the center of the phantom. Each stone was placed within the water-filled lumen of the phantom, and all stones were scanned separately with the Somatom Definition Dual Source CT (Siemens Medical Solutions, Inc., Malvern, Pennsylvania) at a dual kilovolts setting (140 kV with 110 effective mAs, and 80 kV with 468 effective mAs, settings that are fixed by the manufacturer). Detector collimation was 64 × 1.5 mm, and the data were reconstructed at 1.5 mm slice thickness at 0.7 mm intervals. The image data were reconstructed with the body soft tissue algorithm with a 512 × 512 matrix. The field of view for each scan was 25.0 × 25.0 cm.

For image analysis, the datasets were sent to the computer workstation (Leonardo CT Workstation, Siemens Medical Solutions, Inc., Malvern, Pennsylvania). The image analysis of each stone was performed separately and with magnification (Figure 1). The HU of each stone was measured by 3 different methods: (1) Drawing a region of interest (ROI) with a soft tissue window (window:410, level:10) by placing the largest possible circle ROI that still left approximately 0.5 mm at the boundary of each stone on the axial slice in which it was seen at its largest diameter. The highest pixel density of each stone at 80 kV and 140 kV were also recorded by this method; (2) Drawing a ROI by hand that left approximately 0.5 mm at the boundary of each stone on a multiplanar reconstruction image in which the stone was seen at its largest area (Figure 1a). The largest diameter of each stone was also measured at the same slice using an electronic caliper measuring tool (Figure 1b); (3) For CaP and CaOx stones, after defining a volume of each stone by setting the upper HU (<100 HU above the highest pixel density of each stone) and lower HU threshold (25% of the highest pixel density), the average HU of the selected volume of each stone was obtained (Figure 1c-d). For three CaP stones and four CaOx stone, lower thresholds of 27 to 32% of highest pixel density were used instead of 25% to exclude background phantom density. This method was not used for UA stone because the highest pixel densities of UA stones were lower than CaP and CaOx stones, and the lower HU threshold could not separate stone and background phantom density. For measurement using the methods (1) and (2), the HU of each stone at 80kV and 140 kV were obtained using the same ROI. For measurement using the method (3), upper and lower HU threshold had to be set separately for 80kV and 140 KV datasets, and the selected volume for each stone at 80kV and 140 kV were not identical. The HU of phantom using ROI were 160.5 (SD 1.0) at 140 kV and 305.4 (SD 1.6) at 80kV, and were not significantly different for each scan. The HU differences between the two energy sources were calculated as previously described by Mostafavi and associates9 and Mitcheson and associates7 in their previously reported work detailing the analysis of renal calculi with multiple different CT energy levels.

Figure 1. CaOx stone. (A) ROI drawn by hand that left approximately 0.5 mm at the boundary of the stone on a coronal reconstruction image in which the stone was seen at its largest area; (B) The largest diameter of each stone was measured on a coronal reconstruction image in which the stone was seen at its largest area; (C and D) The volume of the stone was defined by setting the upper HU (1600 HU) and lower HU (25% of highest pixel density; 397 HU) thresholds, which provided the average HU of the selected volume (shown as colored voxel).

Statistical analysis was performed with the software Stata IC 10 (StataCorp, College Station, Texas). The three groups were compared with analysis of variance (ANOVA), and a post-hoc comparison of mean values was performed with the Tukey-Kramer HSD test, a single step multiple comparison procedure.


A total of 10 CaOx, 8 CaP, and 8 UA stones were evaluated. The mean stone size for the CaOx group was 4.19 mm, for the CaP group was 5.57 mm, and for the UA group was 4.71 mm. There was no significant difference in stone size between the CaP and CaOx stones (4.19 vs. 5.57, P = 0.119). The UA stone size was also not significantly different from the calcium stones (P = 0.401).

The mean HU for CaP, CaOx, and UA stones are presented as Table 1. The HU of the phantom using ROI were 160.5 (SD 1.0) at 140 kV and 305.4 (SD 1.6) at 80 kV, and were not significantly different for each scan. Table 2 presents the differences between the Hounsfield units obtained by the 80 kV energy source and those obtained by the 140 kV energy source, as well as the ratios of the values obtained by the 80kV source to the values obtained by the 140 kV source. For the three different measurement techniques, the values calculated for each stone type were significantly different from one another. The differences between, and the ratios of, the HUs at 80kV and 140 kV were significantly higher for CaP than CaOx stones (P < 0.05 in all cases).


Few studies have been able to distinguish among stones beyond those composed of UA or non-specific calcium salt. Zarse and associates imaged calculi with a micro CT device, a high resolution non-clinical laboratory instrument that can demonstrate the internal structure of a stone in microscopic detail, to define specific regions of interest within the stone that could then be analyzed in a more targeted fashion with helical CT.11 In this way, helical CT was focused to these areas of interest, and CaOx mineral components were distinguished from CaP mineral components. Mostafavi and associates, who evaluated 6 different types of stones, was able to differentiate UA, struvite and CaOx stones using absolute HUs at 120 kV, but found it impossible to differentiate CaOx from CaP and struvite from cystine at 120 kV.9 However, scanning stones repeatedly at different x-ray tube voltage levels found that the differences between the HUs at 80kV and 120 kV with 240 mA were significantly higher for CaOx stones than for CaP stones. Mitcheson and associates were able to differentiate cystine stones and UA stones, and struvite and UA stones using difference of HUs measured at different x-ray energies (77 kVp with 747 mAs, and 125 kVp with 460 mAs).7 However, the authors were unable to differentiate CaOx and CaP stones, as these stones had more than 5% of pixels with CT values higher than the highest HU that could be displayed by their CT scanner.

One reason that previous studies were unable to successfully differentiate among different calcium salts may be due to the limitations of standard helical CT imaging. A single source CT scanner images patients with a single x-ray source that functions at a single, specific energy. Dual source imaging utilizes two CT tubes that produce two different x-ray energies, 140 kV and 80 kV, which arise from two x-ray sources that are placed at 90 degree angles to one another within the CT. Opposite each x-ray source is a standard detector array. Therefore, the data from two different x-ray sources are acquired at the same time, and the data from each x-ray source are perfectly aligned. The dual source CT workstation allows manipulation of the radiographic dataset, enabling the investigator to measure the HUs of an identical ROI on a single image that was created with both 80kV and 140 kV x-ray energy sources. Scanning the stones multiple times with different x-ray energies, as would be required with a single source CT, can make it difficult to eliminate cumulative measurement errors, as an ROI must be re-drawn on each image scan. In addition, in a clinical setting, an added benefit of dual source CT imaging is the elimination of artifactual effects due to a patient's motion.

The interplay of the two radiographic sources of the dual source CT may detect unique absorption characteristics of the constituent parts of the subject being scanned, such as bone, tissue, stone, and may permit subsequent analysis with techniques such as digital subtraction. Each energy source of the dual source scanner provides its own unique measure of the kidney stone's attenuation values, and most of the interactions result from Compton scattering at the high kilovolt level typically used for CT.13 Our findings with dual source CT are similar to those of Mostafavi, who used repeated single source CT scanning at multiple energy levels, as we both found that the difference in HU was more dramatic for calcium-containing calculi than for noncalcified calculi.9 As described by Mostafavi and associates,9 as well as Mitcheson and associates,7 we, too, used the difference of HUs detected by the two energy sources to capture our significant findings. Interestingly, Mostafavi and associates found that the difference in HU at the two energy levels was greater for calcium oxalate monohydrate than for calcium phosphate, and we found the opposite. This incongruity may be due to either the fact that our calcium phosphate stones were hydroxyapatite, whereas those of Mostafavi were brushite, or it may be due to the fact that our dual source CT scanner used energy levels of 140 kV and 80 kV, and Mostafavi's used a single source scanner at energy levels of 120 kV and 80 kV. Although stone size may affect the results of attenuation measurement through the phenomenon of volume averaging, we found no significant difference in stone size among the UA, CaOx, and CaP groups, suggesting that this was not a confounding factor.

A number of methods to measure attenuation of urinary stones on CT have been described: the highest pixel density within the ROI created overlying the calculus6; the average of 3 different 1 pixel ROI HU measurements obtained from each stone9; placing the ROI over the entire calculus on the slice in which it was seen at its largest diameter14; drawing circular ROI 1 to 2 mm inside of each stone's boundary7; drawing ROI by hand and setting the display level to 75% of maximum attenuation value.10 In our study, we analyzed the subject stones with a relatively large field of view, which closely approximates a clinical protocol. Others have reported that most urinary stones are heterogeneous in density, as characterized on helical CT histograms as well as micro CT imaging.7; 8 ; 11 With this foreknowledge, we elected to use three different methods to characterize our subject stones in order to maximize our ability to obtain accurate HU measurements. We found similar results with all three methods, and by all three measurement techniques there was a significant difference between CaOx and CaP stones, in both HU differences as well as HU ratios.


Although standard clinical helical CT imaging protocols have been able to differentiate UA stones from calcium stones, something that can also be done with plain radiography, these protocols have been unable to reliably differentiate among calcium salts. Our work represents an initial investigation into the ability of dual source CT imaging to define stone composition. In our ex vivo analysis, dual source CT technology can reliably differentiate not just UA stones from calciferous stones, but also hydroxyapatite stones from CaOx stones. The logical follow-up to the present work is to expand our preliminary report, and to examine the ability of dual source imaging to characterize stone composition in living, human stone formers. Further efforts may then be devoted to defining the role of dual source CT imaging in clinical practice, whether it be prescribing appropriate medical therapy or even predicting stone fragility.


We thank the following people for their assistance: Jefferson Scott Graves, RT(R)(CT), Emile Thomas Averill, BS, RT(R)(CT), and William F. Vandaniker, RT(R).


1. A. Heidenreich, F. Desgrandschamps, F. Terrier
Modern approach of diagnosis and management of acute flank pain: Review of all imaging modalities
Eur Urol, 41 (2002), pp. 351–362

2. A. Demirel, S. Suma
The efficacy of non-contrast helical computed tomography in the prediction of urinary stone composition in vivo
J Int Med Res, 31 (2003), pp. 1–5

3. M. Kuwahara, S. Kageyama, S. Kurosu, et al.
Computed tomography and composition of renal calculi
Urol Res, 12 (1984), pp. 111–113

4. G. Pareek, N.A. Armenakas, J.A. Fracchia
Hounsfield units on computerized tomography predict stone-free rates after extracorporeal shock wave lithotripsy
J Urol, 169 (2003), pp. 1679–1681

5. G. Motley, N. Dalrymple, C. Keesling, et al.
Hounsfield unit density in the determination of urinary stone composition
Urology, 58 (2001), pp. 170–173

6. S.Y. Nakada, D.G. Hoff, S. Attai, et al.
Determination of stone composition by noncontrast spiral computed tomography in the clinical setting
Urology, 55 (2000), pp. 816–819

7. H.D. Mitcheson, R.G. Zamenhof, M.S. Bankoff, et al.
Determination of the chemical composition of urinary calculi by computerized tomography
J Urol, 130 (1983), pp. 814–819

8. J.H. Newhouse, E.L. Prien, E.S. Amis Jr, et al.
Computed tomographic analysis of urinary calculi
AJR Am J Roentgenol, 142 (1984), pp. 545–548

9. M.R. Mostafavi, R.D. Ernst, B. Saltzman
Accurate determination of chemical composition of urinary calculi by spiral computerized tomography
J Urol, 159 (1998), pp. 673–675

10. K.C. Saw, J.A. McAteer, A.G. Monga, et al.
Helical CT of urinary calculi: Effect of stone composition, stone size, and scan collimation
AJR Am J Roentgenol, 175 (2000), pp. 329–332

11. C.A. Zarse, J.A. McAteer, M. Tann, et al.
Helical computed tomography accurately reports urinary stone composition using attenuation values: In vitro verification using high-resolution micro-computed tomography calibrated to Fourier transform infrared microspectroscopy
Urology, 63 (2004), pp. 828–833

12. T.G. Flohr, C.H. McCollough, H. Bruder, et al.
First performance evaluation of a dual-source CT (DSCT) system
Eur Radiol, 16 (2006), pp. 256–268

13. M.H. McKetty
The AAPM/RSNA physics tutorial for residents: X-ray attenuation
Radiographics, 18 (1998), pp. 151–163 [quiz 149]

14. M.F. Bellin, R. Renard-Penna, P. Conort, et al.
Helical CT evaluation of the chemical composition of urinary tract calculi with a discriminant analysis of CT-attenuation values and density
Eur Radiol, 14 (2004), pp. 2134–2140

Reprint requests: Brian R. Matlaga, M.D., M.P.H., James Buchanan Brady Urological Institute, The Johns Hopkins University School of Medicine, 600 North Wolfe St, Baltimore, MD 21209

© 1999-2019 Elliot K. Fishman, MD, FACR. All rights reserved.