| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Developments |
1 From the Southern X-ray Clinics (K.A.M., M.A.F.), and Wesley Research Institute (M.R.G., M.A.F.), Wesley Hospital, 451 Coronation Dr, 2nd Fl, Day Center, Auchenflower, Queensland 4066, Australia; and Center for Medical and Health Physics, School of Physical Sciences, Queensland University of Technology, Brisbane, Australia (K.A.M., M.R.G.). Received August 10, 2000; revision requested October 3; revision received January 5, 2001; accepted January 31. Address correspondence to K.A.M. (e-mail: kenmiles@ozemail.com.au).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMaterials and MethodsResults DiscussionAppendix REFERENCES |
|---|
Index terms: Computed tomography (CT), comparative studies, 60.12119 • Computed tomography (CT), perfusion study, 60.12119, 60.12163 • Lung, CT, 60.12112, 60.12115 • Lung, nodule, 60.281 • Lung neoplasms, CT, 60.12112, 60.12115 • Lung neoplasms, diagnosis, 60.30 • Lung neoplasms, PET, 60.12163, 60.30 • Positron emission tomography (PET), comparative studies, 60.12163
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResults DiscussionAppendix REFERENCES |
|---|
Because of the variety of techniques used, comparison the results from different studies has been difficult. The purposes of our study were to evaluate a system of standardization of contrast enhancement measurements that allow comparison between techniques and to illustrate its application to the quantification of enhancement and determination of perfusion within lung nodules.
|
Materials and Methods |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResults DiscussionAppendix REFERENCES |
|---|
Standardized Enhancement Value
In this system, enhancement values were normalized to the enhancement that would be expected if the contrast medium were diluted evenly throughout the patient volume. Since enhancement is determined by the concentration of contrast medium, the mean whole-body enhancement is the dose of contrast medium divided by the volume of the patient. Additional information regarding the standardized enhancement value (SEV) is in the Appendix.
Standardized Perfusion Value
The method for deriving SEV as presented in the Appendix allows standardization between different methods for deriving simple enhancement measures, but it does not enable comparison with dedicated perfusion measurements. This can be achieved with use of the standardized perfusion value (SPV), which relates tissue perfusion to average whole-body perfusion. SPV is defined and further considered in the Appendix.
Conversion of Iodine Concentration to CT Attenuation Value
Calculation of SEV and SPV requires a factor that converts iodine concentration (in milligrams per milliliter) to CT attenuation (in Hounsfield units). To derive this factor, a known volume of contrast medium (iopamidol, Isovue; Bracco, Milan, Italy) in a concentration of 370 mg/mL was diluted to provide eight samples with concentrations of 24.67, 8.22, 2.69, 0.90, 0.30, 0.10, 0.03, and 0.01 mg/mL. This range was chosen to simulate the Hounsfield unit values observed in abdominal CT perfusion studies obtained with 50 mL of 370 mg/mL contrast medium when injected at 7 mL/sec.
The diluted solutions were placed in plastic specimen jars, which were assembled in a light timber mounting. The jars were scanned in air and immersed in a 20-cm-diameter water-filled cylindric phantom. The solutions were also scanned in an anthropomorphic chest phantom. All scanning was performed with our standard clinical perfusion protocol (CT Twin; Elscint, Haifa, Israel). The acquisition parameters were 300 mAs, 120 kVp, 10-mm section thickness, 1 second per section, 360° revolution per section, and 3 seconds between images. Two hundred fifty-six–pixel images were reconstructed by means of standard protocols with a smoothing filter applied.
Hounsfield unit values were obtained from a centrally located region of interest that was two-thirds the size of the sample diameter placed by one author (M.R.G.). The slope of a plot of Hounsfield units against milligrams per milliliter of iodine gave the following calibration factors for this CT scanner with these acquisition parameters: In air, 26.47 HU per milliliter per milligram; in the water-filled phantom, 22.65 HU/mL/mg; and in the chest phantom, 23.55 HU/mL/mg. Correlation coefficients for all graphs were greater than 0.99 (Fig 1). The value of 23.55 HU/mL/mg was used to calculate SEV and SPV for the studies of patients with chest lesions in this study.
|
The CT perfusion study comprised a single-location dynamic sequence of images localized at the level of the largest transverse dimension of the lesion. Iopamidol (50 mL of 370 mg/mL concentration) was administered at 7 mL/sec, and the patients were instructed to hold their breath during the 30-second examination. Data acquisition started at the time of contrast material injection, and 10 1-second images were obtained with a cycle time of 3 seconds, with the same acquisition parameters as were used for the calibration studies.
The CT images were exported by means of the digital imaging and communications in medicine, or DICOM, protocol to a personal computer–based program developed to determine CT perfusion (WINFUN; Department of Radiology, University of Cambridge, England). The following algorithm was used in this software: Perfusion is the maximal slope of the tissue time-attenuation curve divided by the arterial input maximum enhancement.
The aorta, rather than the pulmonary artery, was used as the arterial input because primary pulmonary malignancies generally derive their blood supply from the bronchial circulation. The SPV was determined as shown in the Appendix.
A density of 1 mg/mL was used for all the lesions. All patients received the same dose of contrast agent, 50 mL of 370 mg/mL concentration, which led to a dose of 50 mL x 370 mg/mL x 23.55 HU/mL/mg = 435,675 HU/mL. Pixel-by-pixel analysis allowed generation of parametric maps of SPV (Fig 2).
|
SPV and SUV measurements from the lung lesions were compared by means of standard regression analysis. A bivariate analysis incorporating the patient weight was also performed to determine whether patient weight contributed significantly to any correlation between SPV and SUV.
Comparison with a Previous Multicenter Study of Lung Nodule Enhancement
The ability of the SEV and SPV to enable comparison between different techniques for quantifying contrast enhancement was illustrated further by analyzing the results of a recently reported multicenter study of lung nodule enhancement (6). The multicenter study compared the enhancement values obtained from benign and malignant nodules with a dose of contrast medium standardized for patient weight (420 mg per kilogram of body weight). This dose can be converted to Hounsfield units by using the same calibration factor that was used for our patient series, because the tube voltage for image acquisition was the same in both cases (ie, 120 kVp). Thus, the enhancement (E) values from this study can be converted to SEV as follows (assuming a tissue density of 1g/mL): SEV = E x (1,000 g/[420 mg · 23.55 HU/mL/mg · 1 g/mL]) = E x 0.1011.
|
Results |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResults DiscussionAppendix REFERENCES |
|---|
|
The threshold value for diagnosis of malignancy proposed in the multicenter study was 15 HU, which is equivalent to SEV (or SPV) values of 1.52. On the basis of regression analysis of SPV against SUV (Fig 3), this threshold value was equivalent to an SUV of 2.0, which is close to the value of 2.5 commonly used to distinguish benign and malignant lung nodules at FDG PET. The SPV of 0.8 for the single benign noninflammatory lesion in our series also was below this threshold.
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResults DiscussionAppendix REFERENCES |
|---|
The SPV may also have advantages over absolute perfusion measurements in the assessment of tumors where the basic pathologic process that results in increased contrast enhancement is an increase in the density of microvessels (12). Although a greater density of microvessels will be associated with increased perfusion (blood flow per mass of tissue), perfusion within a vascular bed with a particular microvessel density can also be increased by simply raising the cardiac output. A correction for patient weight is also required, because a higher cardiac output would be distributed throughout a greater volume in a larger patient, with no change in delivery of blood to a given tissue. The SPV corrects for cardiac output normalized for patient weight and could therefore correlate more closely to microvessel density than would perfusion measurements.
The physiologic models on which the SPV is based are valid for organs with relatively long vascular transit times with use of small rapid boluses of contrast medium. Results in previous CT perfusion studies (1–4,7–10) indicate that use of 25–50-mL boluses at 5–10 mL/sec will be satisfactory for most organs. However, washout of contrast medium could result in underestimation of tissue perfusion for organs with short vascular transit times or in use of larger slower bolus injections of contrast media, such as those used for many clinical CT protocols. On the other hand, washout effects are reduced in tissues that demonstrate increased capillary permeability, because a greater proportion of the contrast medium is retained in the extravascular space. Since tumors generally demonstrate increased capillary permeability, the effect of washout from malignant tissue is likely to be small in most cases (12). This is reflected by the similar SPVs obtained for lung nodules in our series and in the multicenter study (6), despite the use of different bolus sizes and injection rates.
The SPV is conceptually similar in its derivation to the SUV used to quantify FDG uptake at PET. The SUV is used to compare FDG uptake in the tissue of interest with the average uptake throughout the body. Likewise, the SPV is used to compare tissue perfusion with average whole-body perfusion. Although tumor perfusion and glucose metabolism are different physiologic processes, recent advances in tumor molecular genetics provide a biologic rationale for an association between these two aspects of tumor physiology. The p53 oncogene, which is frequently expressed in many malignant tumors, including lung cancer (13), is known to promote tumor angiogenesis (14–16) and glucose metabolism (17). The increased microvessel density that results from angiogenesis leads to increased SPV, whereas the increased glucose metabolism produces increased FDG uptake at PET. These conceptual and biologic relationships between SPV and SUV are reflected by the correlation found between these values in our series of lung nodules. The threshold SPV of 1.52 for malignancy that we derived from the multicenter study of lung nodules (6) is equivalent to an SUV of 2.0, which is close to the threshold of 2.5 commonly used in FDG PET (18)
These similarities between SPV and SUV suggest that the CT-derived SPV may be useful, not only for distinction of benign and malignant lesions, but also for acquisition of prognostic information and assessment of treatment response, in a manner similar to current and emerging applications of FDG PET (18). Most patients already undergo CT for investigation of pulmonary nodules, and an SPV measurement could be readily incorporated into the examination. The greater accessibility and lower cost of CT as compared with PET indicates an important advantage for CT with SPV measurement, particularly if findings of our experience are confirmed in larger series. However, it should be noted that staging of lung cancer is likely to be more effectively performed with PET because of its whole-body imaging capability. The comparability of CT measurements of SPV- and PET-derived SUVs for tumors in other body regions, such as the brain, abdomen, and pelvis, also requires evaluation.
Small nodules may create some difficulty for evaluation of contrast enhancement at CT owing to partial volume effects and patient motion, and it is likely at this stage that FDG PET with use of a dedicated system is more effective than perfusion CT for evaluation of small lung lesions. The difficulties arising from patient movement could be addressed by means of respiratory gating or a multisection technique in which the nodule could be identified at a higher or lower section level in the event that it moves from the reference section. Beam-hardening artifacts from high-density venous contrast medium further constrain the applicability of perfusion CT in the thorax, particularly for nodules close to the mediastinum. These artifacts could be reduced by using more dilute contrast medium, but this would result in less nodular enhancement and therefore a lower signal-to-noise ratio.
Conventional SUV measurements have been shown to be body- weight dependent (19). This dependence is probably due to the relatively higher percentage of fat found in heavier people, which contributes to body weight but accumulates very little FDG. Methods that substitute lean body mass or body surface area for body weight have been proposed as weight-independent alternatives (19). Fat also demonstrates low levels of perfusion; hence, SPVs calculated on the basis of lean body mass or surface area might have similar advantages, but this area needs further study. The weight dependence of SUV and SPV calculations could potentially contribute to the correlation between these measurements in our study. This contribution is likely to be negligible, since only weak weight dependence was found for SPV and SUV calculations in our patient group.
In summary, the SPV is an easily derived, physiologically relevant measure of contrast enhancement that allows comparison between various CT techniques and with FDG PET. The SPV also has some theoretic benefits over simple measures of contrast enhancement and dedicated CT perfusion values, particularly in the assessment of tumors. The biologic association of tumor angiogenesis and glucose metabolism and the correlation between CT measurements of SPV and SUV measurements at FDG PET suggest the potential for quantitative contrast-enhanced CT to provide functional information similar to that provided with PET in selected cases. However, this potential needs to be confirmed with further studies in a variety of clinical settings.
|
Appendix |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResults DiscussionAppendix REFERENCES |
|---|
) for humans:
Observed tissue enhancement is scaled by dose to correct for the greater enhancement that occurs with larger doses. The reduced enhancement that would occur in larger patients is compensated by the patient’s weight. The proposed corrections are also consistent with circulatory physiology. The density term is constant for all patients and allows the SEV to be a dimensionless quantity. If the density is assumed to be 1 g/mL, then the patient mass (in grams) will equal the volume (in milliliters). Since enhancement will change with time, the time (in seconds) of the measurement after injection of contrast material should be recorded as a subscript (eg, SEV25 = 6).
The dose of contrast medium should be given in Hounsfield units, which can be calculated with a correction factor that relates iodine concentration (in milligrams per milliliter) to attenuation (in Hounsfield units). This calibration correction factor can be determined for a particular CT scanner and scanning parameters by using a series of samples of diluted iodine contrast medium (see Materials and Methods). Typically, at 120 kVp, an iodine concentration of 1 mg/mL is equivalent to approximately 25 HU of attenuation, whereas at 80 kVp, a concentration of 1 mg/mL is equivalent to 32 HU of attenuation. Thus, at 120 kVp, a concentration of 300 mg/mL of iodine is equivalent to 7,500 HU of attenuation. This value is then multiplied by the volume of the contrast agent administered to give the dose (in Hounsfield units per milliliter). For example, 100 mL of 300 mg/mL contrast agent is equivalent to a dose of 100 mL x 300 mg/mL x 25 HU/mg/mL = 750,000 HU/mL (ie, the Hounsfield unit value if all the iodine was concentrated in 1 mL).
By way of further illustration, a 75-kg (75,000-g) patient, with density of 1 g/mL, given 100 mL of 300 mg/mL contrast agent with an observed organ enhancement of 50 HU at 25 seconds after injection would give an SEV as follows: SEV25 = (50 HU x 75,000 g)/(750,000 HU/mL x 1 g/mL) = 5.0.
Calculation of SPV
The SPV can be defined as
If, for the purposes of illustration, a 70-kg (70,000-g) man had a cardiac output of 7,000 mL/min, the mean whole-body perfusion would be 0.1 mL/min/g. Thus, SPV measurements would be 10 times tissue perfusion measurements.
Substitution of Equation (3) into Equation (2) gives
Under certain circumstances, perfusion can be determined at CT with the following equation, which is based on the Fick principle (20):
t is the density of the tissue of interest. The peak enhancement is determined per pixel or for a region of interest, both of which are volumes. To obtain perfusion in milliliters per minute per gram, t is required.
The cardiac output can also be determined at CT, with the following equation (21):
By substituting Equations (5) and (6) into Equation (3), the SPV can be determined from
Equation (7) reduces to
If t is considered to be 1 g/mL, then this term can be ignored. Alternatively, a more accurate value of 1.05 g/mL could be used.
There is a clear similarity between Equation (1), which defines SEV, and Equation (8), which defines SPV. This similarity provides a physiologic justification for the definition of SEV. Furthermore, it can be seen that within the first pass of the contrast medium, the peak value of SEV (SEVP) will be equal to SPV:
Hence, the simple measurements of peak enhancement can be related to measurements of perfusion. Measurements of enhancement other than the peak value can not be directly related to SPV but can be understood in relation to their anticipated peak enhancement as indicated by the SPV. In addition to Equation (5), other methods are often used to calculate perfusion from dynamic CT data, including rate of enhancement (the slope of the tissue time-attenuation curve) (10) and deconvolution (8). These methods require acquisition of vascular time-attenuation data from which AUC*, the area under the arterial time-attenuation curve corrected for recirculation, can be determined and the SPV calculated with Equation (10), which is obtained by substituting Equation (6) into Equation (4). SPVs derived from differing perfusion calculation methods could be denoted as SPVslope, SPVdecon, and so on.
To illustrate the relationship between perfusion, SPV, and peak enhancement, let us consider splenic enhancement in an 80-kg man with cardiac output of 7,000 mL/min. Splenic perfusion measured at CT is typically 1.4 mL/min/g (10). The SPV for the spleen in this patient would be (1.4 x 80,000)/7,000 = 16, with Equation (4). The same SPV would also be given by a peak splenic enhancement of 75 HU after administration of 50 mL of 300 mg/mL contrast agent (assuming a tissue density of 1 g/mL). D = 50 x 300 x 25 = 375,000 HU/mL for a 120-keV scanning protocol, and SPV = (75 x 80,000)/(375,000 x 1) = 16, with Equation (8).
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, K.A.M.; study concepts and design, K.A.M., M.R.G.; literature research, M.A.F.; clinical studies, M.A.F.; data acquisition, M.A.F.; data analysis/interpretation, M.A.F., M.R.G., K.A.M.; statistical analysis, M.A.F., M.R.G.; manuscript preparation and definition of intellectual content, K.A.M., M.A.F., M.R.G.; manuscript editing, M.A.F.; manuscript revision/review, K.A.M., M.A.F., M.R.G.; manuscript final version approval, K.A.M., M.A.F., M.R.G.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResults DiscussionAppendix REFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  P S MURPHY, T J McCARTHY, and A S K DZIK-JURASZ The role of clinical imaging in oncological drug development Br. J. Radiol., September 1, 2008; 81(969): 685 - 692. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Q. S. Ng, V. Goh, E. Klotz, H. Fichte, M. I. Saunders, P. J. Hoskin, and A. R. Padhani Quantitative Assessment of Lung Cancer Perfusion Using MDCT: Does Measurement Reproducibility Improve with Greater Tumor Volume Coverage? Am. J. Roentgenol., October 1, 2006; 187(4): 1079 - 1084. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Q.-S. Ng, V. Goh, H. Fichte, E. Klotz, P. Fernie, M. I. Saunders, P. J. Hoskin, and A. R. Padhani Lung Cancer Perfusion at Multi-Detector Row CT: Reproducibility of Whole Tumor Quantitative Measurements Radiology, May 1, 2006; 239(2): 547 - 553. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Z. Rumboldt, R. Al-Okaili, and J. P. Deveikis Perfusion CT for Head and Neck Tumors: Pilot Study AJNR Am. J. Neuroradiol., May 1, 2005; 26(5): 1178 - 1185. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K A Miles Perfusion CT for the assessment of tumour vascularity: which protocol? Br. J. Radiol., December 1, 2003; 76(suppl_1): S36 - S42. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K A Miles and M R Griffiths Perfusion CT: a worthwhile enhancement? Br. J. Radiol., April 1, 2003; 76(904): 220 - 231. [Full Text] [PDF]
|
|
|
|
|
|
M A Fuentes, C J Keith, M Griffiths, G Durbridge, and K A Miles Hepatic haemodynamics: interrelationships between contrast enhancement and perfusion on CT and Doppler perfusion indices Br. J. Radiol., January 1, 2002; 75(889): 17 - 23. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
), a water-filled phantom (
), and a chest phantom (
). The calibration factors for the samples are 26.47, 22.65, and 23.55 HU/mL/mg, respectively.
) (r = 0.80, P < .01). Data for the one patient with an organizing pneumonia (