| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Experimental Studies |
1 From the Departments of Radiology (H.J.P.) and Urology (R.A.S., L.G.F.) and the Clinical Research Program (L.A.K.), Children's Hospital Boston, Harvard Medical School, 300 Longwood Ave, Boston, MA 02115; Department of Radiology, University Hospital, Innsbruck, Austria (F.F.); and Department of Radiology, Thomas Jefferson University, Philadelphia, Pa (P.L.O.). Received February 7, 2005; revision requested April 6; revision received July 20; accepted August 25; final version accepted September 12. Supported in part by National Institutes of Health grant 1 R03 EB00543-01 and by Bristol-Myers Squibb Medical Imaging grant CG#21129. Address correspondence to H.J.P. (e-mail: harriet.paltiel{at}childrens.harvard.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXADVANCES IN KNOWLEDGEReferences |
|---|
Materials and Methods: Institutional animal care committee approval was obtained. After 35 rabbits underwent unilateral spermatic cord occlusion, testicular Doppler US and contrast material–enhanced PI imaging were performed. Enhancement data yielded perfusion measurements including mean value during the first 10 seconds, mean value over entire recorded replenishment curve, and curve slope during the first 5 seconds. Calculated perfusion ratios were compared with radiolabeled microsphere–derived perfusion ratios. Two readers assessed testicular perfusion as none, possible, or definite and relative perfusion as greater to the right testis than to the left, greater to the left testis than to the right, or as equal to both testes. With 
statistics, interobserver agreement for all imaging methods was determined. Association between qualitative perfusion categories and radiolabeled microsphere–based perfusion measurements was assessed. Quantitative and qualitative determinations of relative perfusion were compared with radiolabeled microsphere–based measurements.
Results: Correlations between calculated and radiolabeled microsphere–based perfusion ratios were determined (r = 0.49–0.64). Interobserver agreement for presence of perfusion was excellent ( = 0.76), and that for relative perfusion assessment was good ( = 0.55). Neither value varied significantly with imaging method. The percentage of times a testis classified as having definite perfusion had greater perfusion as measured with radiolabeled microspheres than a testis classified as having no perfusion or possible perfusion was higher with PI imaging than with Doppler US (85%–98% vs 72%–89%). Identification of the testis with less perfusion was better with quantitative methods than with qualitative assessment of images by the readers (75%–79% vs 34%–60%, P < .004).
Conclusion: PI imaging, compared with conventional Doppler US methods, provides superior assessment of perfusion in the setting of acute testicular ischemia.
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXADVANCES IN KNOWLEDGEReferences |
|---|
Color Doppler ultrasonography (US) currently is the imaging test of choice in the evaluation of acute scrotal symptoms because of its capability for direct visualization of the testicular blood supply and for documentation of alterations in blood flow without the need for ionizing radiation (4–7). Its use in children, however, often is limited by the difficulty in detection of flow in testes of small volume (8–12). Contrast material–enhanced US techniques that have been developed have resulted in greater sensitivity to flow in small blood vessels (13–20). Harmonic imaging is a US method in which transducers transmit ultrasound at one frequency (the fundamental) and receive signals at twice that frequency (the second harmonic). When harmonic imaging is used in conjunction with microbubble-based US contrast agents, there is greater enhancement of the vascular signal compared with that obtained with conventional contrast-enhanced color Doppler US methods (21,22). With pulse-inversion (PI) imaging, a two-pulse sequence is used with a 180° phase difference to cancel the effect of transmitted second harmonics from nonvascular background tissues on the received signal. This technique results in improved distinction between parenchymal vessels and background tissues (14–18).
The purpose of our study was to quantify relative testicular perfusion with PI imaging in a rabbit model of acute testicular ischemia and to compare PI and Doppler US images in the qualitative assessment of testicular perfusion.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXADVANCES IN KNOWLEDGEReferences |
|---|
Selection of Animals
The study was performed according to a protocol approved by the Animal Care and Use Committee of Children's Hospital Boston, Boston, Mass, and conformed to guidelines issued by the National Institutes of Health for care of laboratory animals. Use of the rabbit in experimental models of testicular ischemia is well established (23–25). Thirty-five adult male New Zealand white rabbits (Millbrook Breeding Labs, Amherst, Mass) with a mean weight of 4.0 kg were examined.
Animal Preparation
General anesthesia was induced by intramuscularly administering atropine sulfate (Baxter, Deerfield, Ill), 0.04 mg per kilogram of body weight, and intravenously administering ketamine (Ketaset; Fort Dodge, Fort Dodge, Ind), 10 mg/kg, and acepromazine maleate (Phoenix Pharmaceutical, St Joseph, Mo), 0.5 mg/kg (Appendix).
Surgical Procedure
A unilateral inguinal incision with exposure of the spermatic cord was performed by one of two urologists (R.A.S., L.G.F.) with both clinical (8 and 19 years, respectively) and large-animal surgical experience (1 year and 13 years, respectively). A right-sided inguinal incision and then a left-sided inguinal incision were performed for consecutive experimental animals. An inflatable vascular occluder with a 6- or 8-mm luminal diameter (OC6, OC8; Kent Scientific, Torrington, Conn) was placed around the spermatic cord, followed by cuff inflation with normal saline to produce, in succession, mild, moderate, and severe degrees of occlusion of the testicular artery and vein (Appendix).
Contrast Agent Administration
The US contrast agent was intermittently infused intravenously by the first author (H.J.P.) at a rate of 1.3 mL/min, with a maximal dose of 20 mL per experimental animal (Appendix). The duration of each infusion was less than 4 minutes.
Image Acquisition
Serial color Doppler, power Doppler, and PI US imaging was performed by the first author (H.J.P., a pediatric radiologist with 16 years of experience) at baseline and after mild, moderate, and severe degrees of spermatic cord occlusion. All studies were performed by using a US unit (model HDI 5000; Philips Medical Systems, Bothell, Wash) and a linear transducer (Philips Medical Systems). Of the 35 experimental animals in the study, 22 were investigated by using a 12–5-MHz linear transducer, and 13 were investigated with a 7–4-MHz linear transducer. Color and power Doppler US images were acquired prior to contrast material administration. PI images were then obtained by using the so-called negative bolus technique, which is based on the destruction-reperfusion principle (Appendix) (26,29,37).
Regional Perfusion Measurements
Reference testicular perfusion measurements were obtained by the first author by using a radiolabeled-microsphere technique (38). At the conclusion of the experiment, each rabbit was sacrificed with an intravenous overdose of pentobarbital (1 mL/4.5 kg). The testes were then removed by one of two individuals (R.A.S. or L.G.F.) and immediately sectioned for regional perfusion determination (Appendix).
Quantitative Image Analysis
Gray-scale testicular parenchymal enhancement during the PI imaging sequences was measured with manufacturer's software (QLAB, version 1.0; Philips Medical Systems), which allows quantification of pixel intensity before logarithmic compression for video display. For each imaging sequence, identical regions of interest (ROIs) were manually placed by the first author over each testis, with four on each side; care was taken to avoid the large vessels located at the testicular periphery. ROIs varied in size from 1.52 to 21.46 mm2 (mean, 7.2 mm2). An effort was made to place corresponding ROIs in the right and left testes at similar depths from the transducer face to avoid differences in signal intensity values caused by differential beam attenuation. Replenishment curves were estimated on the basis of temporal changes in image signal intensity within each ROI.
Qualitative Image Analysis
Image processing.—The digital cine loops were converted to a multimedia file format (Audio Video Interleave; Microsoft, Redmond, Wash) by the first author (H.J.P.) and an administrative assistant with manufacturer's software (QLAB, version 1.0; Philips Medical Systems) and stripped of all identifying information, including date of the study, animal number, and experimental side. Each image file was assigned a randomly generated index number and exported to a password-protected Web site for review. Identifying links to each animal were available only to the first author and to the first author's administrative assistant and were stored in password-protected computer files.
Image review.—Qualitative assessment of all images was performed independently by two readers (F.F., P.L.O.), both radiologists with particular expertise in US (10 and 8 years, respectively), who were blinded as to the side on which the intervention testis and the control testis were located. They classified the perfusion to each testis on every image as either none, possible, or definite. The criterion used to categorize perfusion as possible or definite consisted of a qualitative judgment comparable to that which would ordinarily be employed in clinical practice. They also categorized relative testicular perfusion as right greater than left, right equal to left, or left greater than right (Appendix).
Statistical Analyses
All statistical analyses were performed by one of the authors (L.A.K.). All reported P values were from two-sided tests and were considered to indicate a statistically significant difference when the P value was less than .05.
Quantitative analysis.—Each replenishment curve plotted with data derived from ROI measurements was analyzed separately for the estimation of three measurements of perfusion: (a) The mean value during the first 10 seconds (M10) was used to measure the average perfusion during the first 10 seconds. (The mean total imaging time differed for sequences performed with the lower- and higher-frequency transducers. Use of a common maximum time interval helped to standardize this difference.) (b) The mean value over the entire recorded replenishment curve (Mmax) was used to measure the average perfusion level over the whole curve. Relative to M10, the first few seconds are downweighted in Mmax because they generally represent a smaller percentage of the entire imaging sequence duration. (c) The curve slope during the first 5 seconds (S05) was used to measure how quickly the testis was reperfused. S05 was calculated by fitting a straight line through the data obtained during the first 5 seconds, with the restriction that the line start at (0,0) (Appendix).
Qualitative analysis.—Interobserver agreement about the presence of perfusion and relative perfusion was assessed with the weighted statistic (35). In the analysis of agreement on the presence of perfusion, the control testis and the intervention testis at the baseline occlusion level were classified as having definite perfusion 97% of the time (1251 of 1296 assessments). Therefore, most of this analysis focused on the intervention testis at mild, moderate, and severe levels of occlusion. For the purpose of analysis, relative perfusion of the right versus the left testis was restated as intervention versus control testis (viz, perfusion in the intervention testis less than that in the control testis, equal perfusion in both intervention and control testes, perfusion in the intervention testis greater than that in the control testis). The readers were blinded to which side was the intervention side (Appendix).
Comparison of Quantitative and Qualitative Analyses
For each of the calculated and qualitative methods, we tabulated the percentage of times each calculated method was correct in the identification of the lesser perfused testis by using the radiolabeled microsphere–based measurements as the reference standard (Appendix).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXADVANCES IN KNOWLEDGEReferences |
|---|
|
|
|
|
|
|
|
The readers agreed with each other about the presence of perfusion on the control side in 487 (95%) of 511 imaging sequences, with 97% of the 511 sequences classified as having definite flow. There was excellent interobserver agreement about the presence of perfusion on the intervention side at mild, moderate, and severe occlusion levels (86% agreement; = 0.76; 95% confidence interval: 0.70, 0.83) (Table 5). The PI method with an MI of 0.09 demonstrated the highest interobserver agreement ( = 0.86 vs 0.73–0.74 for the other three methods), although a test of variation across the methods was not significant (P = .39). When statistics were calculated separately according to occlusion level, the highest level of agreement beyond chance occurred with the greatest occlusion level (Table 6).
|
|
|
= 0.55; 95% confidence interval: 0.49, 0.62) (Table 8). The level of agreement was fairly uniform across the imaging methods ( = 0.50–0.59, P = .79).
|
Comparison of Quantitative and Qualitative Analyses
As the ratio of testicular perfusion measured with radiolabeled microspheres decreased (ie, as the difference between the perfusion values of each testis increased), the percentage of correct identification of the less perfused testis increased for all methods. There was generally a large improvement from the perfusion ratio category of 0.60 or more to less than 0.80 to the perfusion ratio category of 0.40 or more to less than 0.60. Therefore, results are presented for all ratios combined and separately for ratios less than 0.60 (Table 9).
|
When the analysis was restricted to cases with the greatest difference in perfusion (low-high perfusion ratios of <0.60), it was easier to identify the less perfused testis, which resulted in a higher percentage of correct assessments. Again, performance of the calculated measurements exceeded that of the qualitative assessments, although not all of the comparisons were significant. The performance of calculated measurements was especially promising, with an MI of 0.09. The best performance of the qualitative assessments was for color Doppler US, with rates of 74% and 84% for the two readers.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXADVANCES IN KNOWLEDGEReferences |
|---|
Limitations of the study included restriction of imaging to a single testicular tissue plane and inclusion of testicular parenchymal vessels of varying size within each ROI.
When testicular perfusion measurements are based on imaging limited to a single tissue plane, one presumes that these measurements are representative of the entire organ. There may, in fact, be substantial in-plane testicular flow variability. There is, therefore, a need for volumetric flow information to more accurately determine perfusion within the testis as a whole. Current scrotal imaging devices do not permit simultaneous acquisition of reperfusion data from multiple tissue planes. It is possible, however, that averaging of data obtained from two or more tissue planes during a short time would result in an improved estimate of total organ perfusion. Data from multiple planes would also yield information on perfusion changes in the direction orthogonal to the image planes and might provide an opportunity for more accurate alignment between the left and right testes for comparison.
Quantitative analysis in this study relied on a totally subjective determination of parenchymal ROIs, with the potential for erroneous inclusion of large feeding vessels and structures adjacent to the testes. Physiologic differences in flow velocity within the macro- and microcirculation, as well as size and position of the ROIs, have been shown to influence tissue replenishment curves (39–41). The exponential function y = A[1 – e–ßt] has been applied to replenishment curves obtained in vivo to assess perfusion of kidney, myocardium, and brain (40,42–45). As noted by Lucidarme et al (39), however, this exponential model is valid only if the concentration of microbubbles that enter the ROI immediately after the destruction pulse is constant. For this to occur, the contrast medium infusion rate must be constant and blood vessels entering the ROI must not have been previously subjected to the US beam. These authors showed that the plateau of tissue enhancement depends on the fractional blood volume in the ROI, the length of the feeding vessels subjected to the US beam before they reach this region, and the flow rate. Potdevin et al (41) demonstrated in phantom experiments that the shape of the replenishment curve is substantially affected by the range of velocities within an ROI. Testicular replenishment curves have not been previously investigated. In many cases, because the assumed parametric model was not well suited to the data, we abandoned it in favor of three estimates of perfusion detailed before, namely M10, Mmax, and S05. In the future, we intend to perform studies that will incorporate replenishment data derived from multiple tissue planes and will investigate a variety of curve-fitting algorithms.
A possible limitation of the study was the use of an occlusive ischemic model as a surrogate for testicular torsion. The degree of ischemia induced by a defined degree of torsion can be variable. We therefore employed a more straightforward method of producing ischemia through direct occlusion of the spermatic cord to investigate the capability of PI imaging to depict perfusion differences between intervention and control testes.
In conclusion, quantitative and qualitative PI imaging, compared with conventional Doppler US methods, provides a superior assessment of perfusion in acute testicular ischemia. The results of our investigation hold promise in regard to the utility of PI imaging for the diagnosis of testicular ischemia in patients with acute scrotal pain and warrant further methodological investigation and refinement.
Practical application: PI imaging is of potential utility in the diagnosis of testicular ischemia in patients with acute scrotal pain, especially in children with small testes in whom conventional Doppler imaging methods provide a suboptimal assessment of both the presence of perfusion and relative perfusion to the symptomatic and contralateral testes.
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXADVANCES IN KNOWLEDGEReferences |
|---|
Surgical Procedure
The volume of saline required to produce mild to severe levels of occlusion varied from animal to animal, with a range of approximately 0.05–1.50 mL. The degree of occlusion was visually determined by the first author in every instance by monitoring flow within the ipsilateral testis with color Doppler US while increasing pressure within the cuff and comparing the resultant flow within this testis to flow within the contralateral control testis. The criteria used to categorize perfusion as mildly, moderately, or severely diminished relative to that in the control testis consisted of a qualitative judgment comparable to that which would ordinarily be employed in clinical practice.
Contrast Agent Administration
The contrast agent used in this study is a perflutren lipid–coated microsphere composed of octafluoropropane encapsulated in an outer lipid shell. The mean range in diameter of the microsphere particles was 1.1–3.3 µm. One milliliter of this contrast agent suspension contains a maximum of 1.2 x 1010 perflutren lipid microspheres. For each experimental animal, 1.3 mL of contrast agent suspension was reconstituted with normal saline (0.9% NaCl solution) to yield a 20-mL solution. The contrast agent solution was placed in a plastic syringe that was connected to the jugular venous catheter by means of an infusion pump (model 55-1111; Harvard Apparatus, South Natick, Mass). Between infusions, the syringe was gently rotated and inverted to resuspend the contrast agent, which had a tendency to settle out of solution over time.
Image Acquisition
A steady infusion of the contrast agent was established, and a US probe was placed over an ROI. Delivery of a burst of high-energy US pulses into the tissues caused microbubble destruction within the imaging plane (26–28). Low-energy US scanning performed immediately after delivery of the high-energy US pulses permitted imaging of a progressive increase in tissue US signal intensity due to replenishment of the microbubbles through arterial perfusion.
We investigated the 12–5-MHz and 7–4-MHz probes since, in our practice, both are employed in the evaluation of the pediatric scrotum. We were interested in comparing the results from data acquired with these two probes because the image resolution and sensitivity to low flow provided by the higher-frequency probe are generally considered superior to those obtainable with the lower-frequency probe, whereas the resonance frequency of the microbubbles used in the contrast agent interacts better with the 7–4-MHz frequencies than with the 12–5-MHz frequencies. This observation theoretically would result in more robust PI imaging data. Imaging was performed with the US probe fixed by means of a mechanical arm over a midtransverse plane of the scrotum to include both testes within the field of view. The same machine settings were used for acquisition of all of the color and power Doppler images, as well as for all of the PI images. One or two focal zones were positioned adjacent to the mid-to-lower portions of the testes. Gray-scale and color gain were adjusted during the course of the color and power Doppler sequences for image optimization. PI imaging was performed by using a 170-dB dynamic range with persistence turned off. Gray-scale gain was adjusted prior to contrast agent administration and was not altered after it.
At baseline and at each level of spermatic cord occlusion, an intravenous infusion of the contrast agent was established, followed by intermittent PI imaging at a low MI setting until the blood concentration of the microbubbles reached equilibrium. Equilibrium was visually determined by the radiologist who performed the US examination and was generally achieved within 90 seconds. Continuous PI imaging at a low MI was then performed for several seconds, followed first by delivery of seven destructive image pulses into the scrotum at maximal power (0.6–0.9 MI) and then by a reversion to low-MI imaging. Images were acquired uniformly over time for an average of 14 seconds (range, 7–20 seconds) during which contrast agent replenishment to a steady-state level within the imaging plane was achieved. The sampling interval was 0.075 seconds in 90% of the imaging sequences, with a range of 0.0375–0.115 seconds. A scanning sequence performed with a low MI setting of 0.16 was immediately followed by a second scanning sequence performed with a low MI setting of 0.09.
Digital cine loops of all imaging sequences recorded from the time of initiation of continuous low-MI imaging through microbubble destruction and subsequent full contrast agent replenishment were transferred to a personal computer for image processing and analysis.
Regional Perfusion Measurements
A different radioisotope was used for each perfusion determination: cerium 141 (baseline), ruthenium 103 (mild occlusion), niobium 95 (moderate occlusion), and scandium 36 (severe occlusion) (PerkinElmer Life and Analytical Sciences, Billerica, Mass). Each of these radioisotopes has a unique photon energy and therefore can be measured independently. For every perfusion measurement, a 0.2-mL solution containing 10 µCi (0.74 MBq) of 15-µm-diameter radiolabeled microspheres was injected into the left ventricular catheter. A reference blood sample was simultaneously removed from the femoral artery with a 5-mL syringe and withdrawal pump (model 55-1143; Harvard Apparatus) at a rate of 2.0 mL/min for 1.25 minutes. The withdrawal interval was measured with an electronic timer.
After sacrifice, each testis was used in its entirety for perfusion measurements. In the first three rabbits, the testis was divided longitudinally into two strips; in the remaining rabbits, the testis was divided into four strips. The tissue and reference blood samples were weighed on a balance (model AB204; Mettler-Toledo, Greifensee, Switzerland). Radioactivity of the tissue and of reference blood samples was measured in a deep-well gamma counter (Packard, Downers Grove, Ill). Regional perfusion measurements (expressed as [mL · g–1]/min of testicular tissue) for each tissue sample were determined. The measurements for the individual tissue strips were averaged to obtain a mean perfusion measurement for each testis. By averaging the perfusion measurements from four tissue samples as compared with two tissue samples, we were able to increase the stability of the data.
Image Review
Readers A and B accessed the password-protected Web site and sequentially downloaded the randomly ordered image files. Each image file could be reviewed repeatedly without penalty and was classified by means of a pop-up menu. Once an image file was electronically classified and submitted, it was no longer available for review and its classification could not be altered.
Intraobserver reliability was evaluated by having each reader assess 40 (8%) of 511 imaging sequences twice. The percentage of times each reader agreed with himself in regard to the classification of the presence of perfusion in each testis (ie, none, possible, or definite) and in regard to the classification of relative perfusion (ie, perfusion of the right testis greater than left, equal perfusion of both testes, or perfusion of the left testis greater than right) were tabulated.
Statistical Analyses
Quantitative analysis.—In the preliminary analysis, a nonlinear curve that has been used in the literature was fit to each ROI, with y = A[1 – e–ßt], where ß represents mean microbubble velocity, t is time in seconds, A represents fractional blood volume, and the product of ß and A represents tissue perfusion (26,29,30). The assumed parametric model was not, in many cases, well suited to the data and did not permit an estimation of model parameters. This approach was not pursued further.
For each PI imaging sequence, perfusion estimates from the four left and four right testicular ROIs were averaged to produce single left and right summary statistics, and ratios of the perfusion estimate for the intervention testis to that for the control testis were calculated. Thus, the ROI data for each PI imaging sequence were reduced to intervention-control testicular perfusion ratios for M10, Mmax, and S05. These calculated intervention-control perfusion ratios derived from the testicular replenishment curves were then compared with the ratios derived from the radiolabeled microsphere–based regional perfusion measurements.
The performance of the calculated measurements of relative testicular perfusion was assessed by determining (a) the nonparametric Spearman correlation between calculated perfusion ratios and radiolabeled microsphere–based perfusion ratios and (b) the mean difference between calculated perfusion ratios and radiolabeled microsphere–based perfusion ratios. Visual examination of the replenishment curves suggested that there was less noise in the data derived from the lower-frequency (7–4-MHz) US transducer. When we tested whether correlations varied significantly according to calculated measurement (M10, Mmax, S05) or MI (0.16 or 0.09), we accounted for the nonindependence induced by estimating different ratios in the same animal (31). This was not necessary when we compared transducer frequencies (12–5-MHz transducer vs 7–4-MHz transducer); results of a test for independent data were sufficient (31). A repeated-measures analysis model with a robust variance estimate was used to assess whether the mean differences were significantly different from zero and whether they varied according to calculated measurement, MI, or transducer frequency (32).
Qualitative analysis.—The value was calculated separately for each imaging method—namely color Doppler imaging, power Doppler imaging, and PI imaging at MI levels of 0.16 and 0.09—transducer frequency, and all imaging methods combined. To investigate whether there might be a graded association with the degree of spermatic cord occlusion, for the intervention testis was calculated separately according to occlusion level. In addition to the baseline, mild, moderate, and severe categorizations, occlusion categories were created on the basis of radiolabeled microsphere–based perfusion measurements of 0.30 or greater, of 0.20 or greater to less than 0.30, of 0.10 or greater to less than 0.20, and less than 0.10 (mL · g–1)/min to analyze the results in terms of more homogeneous perfusion categories.
Qualitative measurements of perfusion were compared with radiolabeled microsphere–based perfusion measurements by calculating the percentage of times a testis classified as having definite flow had greater flow as measured with radiolabeled microspheres than a testis rated as having none or possible flow. This is equivalent to calculating the area under a receiver operating characteristic curve for comparison of radiolabeled-microsphere perfusion measurements in testes classified as having definite perfusion with those in which perfusion was classified as possible or none (33). In contrast to the usual receiver operating characteristic curve setting, the roles of the reference standard and the diagnostic test are reversed here. Confidence intervals for these quantities were constructed by using the bootstrap method (34).
Comparison of Quantitative and Qualitative Analyses
For quantitative assessments, if the side for which a lower calculated measurement of testicular perfusion (ie, M10, Mmax, S05) was the same as the side for which the radiolabeled microsphere–based perfusion measurement was lower, then the calculated measurement was considered correct. For qualitative assessments, if a reader classified perfusion of the left testis as less than that of the right and the perfusion measurement of the left testis was lower or, similarly, if a reader classified perfusion of the right testis as less than that of the left and the perfusion measurement of the right testis was lower, then the reader was considered correct. If a reader classified perfusion as equal to both testes, the reader was considered incorrect regardless of the perfusion measurement. (In fact, for the right and left testes, radiolabeled microsphere–based perfusion measurements never were exactly the same.)
The percentage of correct assessments was determined separately for the six calculated measurements (M10, Mmax, and S05 applied to PI imaging with an MI of 0.16 and to PI imaging with an MI of 0.09) and for the eight qualitative assessments (the four imaging methods of color Doppler US, power Doppler US, PI imaging with an MI of 0.16, and PI imaging with an MI of 0.09, each assessed by two readers). For each combination, the results were further stratified according to the true relative perfusion measurements obtained with radiolabeled microspheres by using four categories that were based on the ratio of lower perfusion to higher perfusion measurements: perfusion ratio category of 0.80 or more to less than 1.00, perfusion ratio category of 0.60 or more to less than 0.80, perfusion ratio category of 0.40 or more to less than 0.60, and perfusion ratio category of zero or more to less than 0.40. Comparison of each calculated measurement with each qualitative measurement was performed by using the McNemar test (35). The Bonferroni-Holm method was used to adjust for multiple comparisons (36).
|
ADVANCES IN KNOWLEDGE |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXADVANCES IN KNOWLEDGEReferences |
|---|
|
FOOTNOTES |
|---|
Abbreviations: M10 = mean value during first 10 seconds • MI = mechanical index • Mmax = mean value over entire recorded replenishment curve • PI = pulse inversion • ROI = region of interest • S05 = curve slope during first 5 seconds
Author contributions: Guarantor of integrity of entire study, H.J.P.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, H.J.P.; experimental studies, H.J.P., R.A.S., F.F., P.L.O., L.G.F.; statistical analysis, L.A.K.; and manuscript editing, H.J.P., L.A.K.
See Materials and Methods for pertinent disclosures.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXADVANCES IN KNOWLEDGEReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS |
