Original Research
Nuclear Medicine and Molecular Imaging
August 2009

Comparison of 99mTc-DMSA Dual-Head SPECT Versus High-Resolution Parallel-Hole Planar Imaging for the Detection of Renal Cortical Defects

Abstract

OBJECTIVE. Renal cortical scintigraphy with 99mTc dimer captosuccinic acid (DMSA) is the standard method to detect acute pyelonephritis and cortical scarring. Different acquisition methods have been used: planar parallel-hole or pinhole collimation and single photon emission tomography (SPECT). Publications support the superiority of each; few comparative studies have been reported, with mixed results. We have compared planar parallel-hole cortical scintigraphy and dual-head SPECT for detection of cortical defects.
MATERIALS AND METHODS. Forty patients (37 children, 3 adults; 9 male, 31 female) were imaged 3 hours after injection of 99mTc-DMSA with dual-head SPECT and planar imaging (posterior, left, and right posterior oblique views with a parallel-hole collimator). For each patient, planar and SPECT images were evaluated at different sittings, in random order, by three independent observers. Twelve cortical segments were scored as normal or reduced uptake. The linear correlation coefficient for the number of abnormal segments detected between readers, techniques, and segments was calculated.
RESUlTS. No significant difference was seen in the average number of abnormal segments detected by planar versus SPECT imaging; 2.1 for planar imaging and 2.2 for SPECT (p = 0.84, two-tailed). For all observers, the average correlation coefficient for SPECT alone, planar imaging alone, and between techniques (SPECT vs planar imaging) was high (r = 0.93–0.94). Applying nonparametric Spearman's rank analysis, the average correlation remained high (r = 0.70–0.75). Correlation between readers, techniques, and segments for methods and readers was also good (r = 0.69–0.77).
CONClUSION. 99mTc-DMSA renal cortical imaging using dual-head SPECT offers no statistically significant diagnostic advantage over planar imaging for detection of cortical defects.

Introduction

Renal cortical scintigraphy with 99mTc dimer captosuccinic acid (DMSA) is the standard method to detect acute pyelonephritis and renal cortical scarring [1]. Different acquisition methodologies have been used, including planar parallel-hole collimator imaging, pinhole collimator imaging, and SPECT. For different scintigraphic studies and various radiopharmaceuticals, SPECT has often proven superior to planar imaging [24]. There are data to support the accuracy of each DMSA imaging method; however, few comparative studies have been published. The results are mixed, and there is no consensus regarding which method is superior [59]. Many of the published studies used SPECT cameras that are much less commonly used today (e.g., triple-head and single-head detector cameras). Many clinics do not use SPECT because it is more technically demanding and requires pediatric sedation for young children. The purpose of our investigation was to directly compare planar cortical scintigraphy and dual-head camera SPECT for detection of renal cortical uptake defects and to determine whether there is a significant diagnostic difference.

Materials and Methods

We investigated 40 consecutive patients (9 male, 31 female) with 78 kidneys presenting to our facility between August 31, 2006, and March 21, 2007. Age range was 1 month to 80 years (mean, 11 years; median, 1 year). All were children except for two 21- and 22-year-old women with recent urinary tract infections and histories of pyelonephritis and one 80-year-old woman referred as part of a preoperative evaluation for renal tumor. All the remaining children were referred for renal cortical imaging because of an initial or recurrent urinary tract infection to evaluate for scarring or pyelonephritis and to calculate differential renal function. This investigation was approved by the Johns Hopkins University institutional review board, and informed consent forms were signed by the patients or their parents or guardians.
Patients were injected with a weight-adjusted dose of 99mTc-DMSA based on the maximal dose of 99mTc-DMSA of 5 mCi (185 MBq). The minimum administered dose was 1 mCi (37 MBq). All children aged 6 months to 4 years were sedated before the study with either oral chloral hydrate or IV pentobarbital. Scintigraphy was started 3 hours after injection. Both planar imaging and SPECT were acquired on all patients using either an Infinia (GE Healthcare) or a Siemens e.cam (Siemens Medical Solutions) dual-head gamma camera equipped with low-energy, high-resolution, parallel-hole collimators.
In children who were sedated, SPECT images were acquired first with the dual-head gamma camera. The imaging field was centered on the kidneys to include from the xiphoid process to the symphysis pubis in a 128 × 128 matrix, and a zoom was applied depending on the size of the patient (maximum 2×). The acquisition parameters for each head used a 180° arc (60 images per head), 30 seconds per image, 3° per rotation. The total acquisition time was 30 minutes. Reconstruction was performed similarly on both cameras using a Hann prefilter (cutoff frequency, 0.9 cm–1; order, 0) and a Butterworth postfilter (cutoff frequency, 0.5 cm–1; order, 10) with two iterations and 10 subsets.
Planar images were acquired immediately after SPECT images in a 256 × 256 matrix. Three projections were obtained: posterior, left posterior oblique, and right posterior oblique, for 5–10 minutes each (approximately 300,000–500,000 counts per image). Images were magnified so that the kidneys would fill two thirds of the field of view. Planar images were acquired first in nonsedated patients.

Image Interpretation

The planar and SPECT scintigrams were reviewed independently by three readers: two second-year nuclear medicine residents and a nuclear medicine physician with 25 years' experience. Each kidney was divided into 12 cortical segments. A segment was defined as abnormal if there was reduced or absent cortical radiotracer activity. No attempt was made to distinguish scarring from infection. Abnormal segments were recorded on a worksheet with the 12 segments for each kidney defined by a schematic drawing (Fig. 1). The total number of abnormal segments for each kidney (0–12) was recorded.
A counterbalanced crossover study design was used for image interpretation. The design was implemented as follows: For each reader, the patients were divided randomly into two equal groups. Initially, each reader reviewed the SPECT data for their first group of patients and the planar data for their second group of patients. After at least 2 weeks, the reader reviewed the remainder of the data—that is, the planar data for their first group of patients and the SPECT data for their second group of patients. Therefore, at random, half of the patients had their planar images interpreted first, and half of the patients had their SPECT images interpreted first. Thus, for a given patient, the two techniques were reviewed at least 2 weeks apart. This design attempts to remove as many sources of bias as possible or to balance their effects.

Data Analysis

The results were analyzed by a separate member of the team. Seventy-eight renal units were evaluated. The number of renal segments labeled abnormal for each patient, by each of the three readers, and for both techniques were compared.
In our analysis, we calculated the linear correlation coefficient (r value) and the average r value of the number of abnormal segments detected between all possible combinations of reader and technique in the following manner: The linear correlation coefficient (r value) for the number of abnormal segments detected between readers was calculated for planar imaging, SPECT, and between the two techniques for all readers. From this data, we calculated the average r value of each of these parameters
In a second analysis, only kidneys in which at least one reader identified both affected and preserved renal parenchyma were included. In our data set, we had 49 renal units that all three readers labeled with zero affected segments, and two that all three readers labeled with 12 affected renal segments. Both groups were excluded. Twenty-seven renal units remained in this “censored” data set. These same analyses were performed on this censored data. Nonparametric analysis (Spearman's rank correlation) was then applied to the censored data, resulting in the nonparametric correlation coefficient and the nonparametric average r value.
To compare the precision of lesion localization with planar imaging and SPECT, a segment-by-segment analysis was performed. Segments were scored 1 if they were marked as affected and 0 if they were marked unaffected. Linear correlation was calculated between the scores assigned to segments on a per-segment basis. For a given method (planar or SPECT), the average correlation coefficient was calculated for all kidneys between all reader combinations, for SPECT–SPECT (different readers) and planar–planar (different readers), and for SPECT–planar (different readers) and SPECT–planar (identical readers).
Fig. 1 Illustration of 12 cortical segments of kidney used for scoring normal and abnormal regions for planar and SPECT imaging. This is same schematic diagram used for on-going Randomized Intervention for Children with Vesicoureteral Reflux multicenter investigation. (Adapted from www.wikidoc.org/index.php/Nephrology)

Results

No significant difference was seen in the average number of abnormal segments detected by planar imaging or SPECT, which were 2.1 for planar imaging and 2.2 for SPECT (p = 0.84, two-tailed).
For planar 99mTc-DMSA imaging, the linear correlation coefficient between readers was 0.92–0.95; for the censored data, it was 0.74–0.83; and applying Spearman's rank analysis, it was 0.66–0.83 (Table 1). For SPECT, the linear correlation coefficient between readers was 0.91–0.95; for the censored data, it was 0.72–0.83; and applying Spearman's rank analysis, it was 0.67–0.76 (Table 2). For planar imaging versus SPECT, the linear correlation coefficient for all readers was 0.90–0.98; for the censored data, it was 0.65–0.93; and applying Spearman's rank analysis, it was 0.62–0.92 (Table 3).
TABLE 1: Correlation Coefficient Between Readers for Planar 99mTc-DMSA Imaging
ReaderLinear Correlation CoefficientLinear Correlation Coefficient (Censored Data)Spearman's Rank Correlation
1 vs 20.950.830.83
1 vs 30.920.740.66
2 vs 3
0.94
0.80
0.77
Note—Censored data are defined in Materials and Methods, Data Analysis. DMSA = dimer captosuccinic acid.
TABLE 2: Correlation Coefficient Between Readers for SPECT 99mTc-DMSA Imaging
ReaderLinear Correlation CoefficientLinear Correlation Coefficient (Censored Data)Spearman's Rank Correlation
1 vs 20.910.720.67
1 vs 30.930.740.67
2 vs 3
0.95
0.83
0.76
Note—Censored data are defined in Materials and Methods, Data Analysis. DMSA = dimer captosuccinic acid.
TABLE 3: Correlation Coefficient Between Readers for Planar Imaging Versus SPECT
PlanarSPECTLinear Correlation CoefficientLinear Correlation Coefficient (Censored Data)Spearman's Rank Correlation
Reader 1Reader 10.900.650.62
Reader 1Reader 20.930.780.76
Reader 1Reader 30.920.740.72
Reader 2Reader 10.920.740.68
Reader 2Reader 20.980.930.92
Reader 2Reader 30.950.860.78
Reader 3Reader 10.910.690.62
Reader 3Reader 20.930.740.72
Reader 3
Reader 3
0.92
0.73
0.72
Note—Censored data are defined in Materials and Methods, Data Analysis.
The average correlation coefficient between observers was 0.94 for planar imaging, 0.93 for SPECT, and 0.93 for SPECT versus planar imaging (Table 4). In the censored data, the average r value between observers was 0.79 for planar imaging, 0.76 for SPECT, and 0.76 for SPECT versus planar (Table 4). Applying nonparametric Spearman's rank analysis, the average r value was 0.75 for planar imaging, 0.70 for SPECT, and 0.73 for SPECT versus planar (Table 4).
TABLE 4: Average Interobserver Correlation for All Readers for Planar Imaging, SPECT, and Planar Imaging vs SPECT
All ReadersLinear Correlation CoefficientLinear Correlation Coefficient (Censored Data)Spearman's Rank Correlation
Planar0.940.790.75
SPECT0.930.760.70
SPECT vs Planar
0.93
0.76
0.73
Note—Censored data are defined in Materials and Methods, Data Analysis.
The results of the segment-by-segment analysis (Table 5 and Fig. 2) show the correlation coefficients for planar–planar (different readers) to be 0.71–0.80 (average, 0.74), for planar–SPECT (same readers) to be 0.69–0.81 (average, 0.74), for planar–SPECT (different readers) to be 0.65–0.75 (average, 0.71), and for SPECT–SPECT (same readers) to be 0.67–0.77 (average, 0.71).
TABLE 5: Results of Segment-by-Segment Analysis
Segment
Comparison123456789101112
P—P0.800.770.710.690.700.770.760.700.710.720.730.78
P—S, same reader0.810.730.710.720.690.710.790.740.740.720.710.77
P—S, different readers0.750.730.690.670.650.730.730.740.720.660.680.73
S—S
0.77
0.76
0.69
0.67
0.69
0.72
0.73
0.69
0.70
0.69
0.70
0.75
Note—Results are represented as average for all kidneys between all reader combinations, for planar—planar (P—P) (different readers); planar—SPECT (P—S), same reader; planar—SPECT, different readers; and SPECT—SPECT (S—S), different readers.

Discussion

Technetium-99m DMSA scintigraphy is a well-established standard method for the detection of acute pyelonephritis and its differentiation from lower tract infection, and for the detection of renal scarring in children with a history of urinary tract infection [1]. Technetium-99m DMSA is more sensitive than urography or ultrasound for detecting cortical defects [10] and performs at least as well as helical CT or MRI without the use of contrast material and the higher radiation dose of CT [11].
Animal studies, using histology as a reference standard, report high sensitivity and high specificity for the diagnosis of pyelonephritis in piglets. Using planar imaging and a converging collimator, Rushton et al. [12] reported 87% sensitivity and 97% specificity. Parkhouse et al. [13] reported a sensitivity of 100% and specificity of 89% using a multipurpose parallel-hole collimator. Using SPECT, Giblin et al. [14] found a sensitivity of 97% and specificity of 93%. The only comparative piglet study was that of Majd et al. [15], who reported 99mTc-DMSA SPECT scintigraphy to be 92% sensitive and 82% specific compared with pinhole imaging, which was 83% sensitive and 95% specific.
Several clinical studies have compared SPECT with planar imaging in humans [59]. In 1990, Tarkington et al. [5] compared triple-detector SPECT with pinhole imaging and found that SPECT “enhanced” diagnostic information in 71% of patients. Those authors reported detection of cortical defects with SPECT in 63% of kidneys that appeared normal on pinhole imaging. Yen et al. [6] found significantly (p = 0.05) more defects using single-detector SPECT than planar imaging. Applegate et al. [7], comparing planar parallel-hole, pinhole, and triple-detector SPECT, found SPECT to have better diagnostic certainty than pinhole (p = 0.031) or planar imaging (p = 0.001) at showing definite defects. However, the total number of kidneys with defects was similar with each scintigraphic technique and not significantly different. Neither method proved superior at showing multiple as opposed to single defects. Other studies have been even less convincing. Mouratidis et al. [8] found more defects detected by SPECT than planar imaging, but these differences were not significant (p = 0.54). Everaert et al. [9] did not detect significantly different numbers of lesions for SPECT or planar imaging.
The two studies showing the most promising results for SPECT, Applegate et al. [7] and Tarkington et al. [5] used triple-detector SPECT. Triple-detector systems with ultra-high-resolution collimators provide better image resolution; however, they are not generally available today. Two-head cameras are commonly used today; however, there have not been direct comparisons with the triple-head systems. Single-head SPECT cameras are still used by some; the advantage of these cameras in renal SPECT is even less clear. Mouratidis et al. [8], using a single-detector camera, found more defects detected by SPECT than by planar imaging; however, their results were not statistically significant. Yen et al. [6] found significantly (p = 0.05) more defects using single-detector SPECT and planar imaging [6]. However, planar images were obtained only in the posterior projection.
Fig. 2 Graph shows Spearman's correlation for individual renal segments imaged with parallel-hole planar imaging and 99mTc dimer captosuccinic acid SPECT. S = same readers, D = different readers.
With clinical studies there is no reference standard. Even though more defects might be detected by SPECT, it is difficult to ascertain whether SPECT detects more true-positives or more false-positives. In the piglet study of Majd et al. [15], comparing dual-head SPECT versus pinhole imaging, the sensitivity of SPECT for the detection of acute pyelonephritis was higher than that of pinhole imaging; however, the specificity was lower, a trade-off for improved sensitivity with an increase in the number of false-positives. The overall accuracy of the two imaging techniques was the same.
For planar imaging, there has been only one direct comparison between pinhole imaging and parallel-hole collimator imaging of which we are aware [7]. That study reported that pinhole imaging and SPECT were more effective at showing definite defects compared with parallel-hole collimator imaging (p = 0.03). However, most centers use parallel-hole rather than pinhole collimators. This was the method that has been used in our clinic as well and thus was used for the comparison in this study.
Because of the diversity of techniques, differing results [16], improvement and change in cameras over time, and the limited statistical analysis of prior studies, we think that the issue of SPECT versus planar imaging has not been resolved. Thus, we initiated an investigation using a counterbalanced crossover design for image interpretation—that is, a statistical method that attempts to remove as many sources of bias as possible by balancing their effects. Because initial readings in one technique can affect the results of a reading in the other technique immediately following, each patient was randomly assigned to the first technique to be read, either SPECT or planar imaging—for example, reader 1 might read SPECT first for a given patient and reader 2 would read planar first. Then the other technique was reviewed for each patient at least 2 weeks later. This crossover approach helps cancel out the effect of the order on interpretation. It also balances the factors that could affect reader effectiveness—energy, concentration level, familiarity with the imaging study, information to be abstracted from the images, and database entry acumen. By dividing every task into two halves and randomizing the order in which we perform them, confounding factors will cancel out on average. This design has been used before—for example, in MRI [17]—but it has not been used previously for comparison of DMSA imaging methods.
Using this vigorous statistical design, our investigation found no significant difference between the average numbers of affected segments per kidney detected using SPECT (n = 2.2) versus planar imaging (n = 2.1). The linear correlation for the number of abnormal segments detected between all possible combinations of reader and technique was very good, greater than 0.91 and 0.93, respectively. Because kidneys with no defects (0 segments involved) or extensive defects (12 segments involved) are likely to have a high correlation between imaging methods and the remaining kidneys are the ones of more pertinent interest, we performed a second analysis using censored data—that is, we excluded the patients with 0 or 12 segments' involvement. The correlation coefficient, although reduced as would be expected, showed that results of planar imaging and SPECT were similar and not significantly different. Because the number of defects in each kidney is a discrete data point, our data could not be assumed to have normal distribution [18]. Therefore, a nonparametric analysis correlation (Spearman's rank) was applied. The correlation continued to be good (Tables 1, 2, 3, 4). Interestingly, no significant difference was seen in interpretation between the nuclear medicine resident physicians and the experienced nuclear medicine physician. Our segment-by-segment analysis also found that the precision in lesion localization is similar between planar imaging and SPECT. This further strengthens our findings that planar imaging is grossly equivalent to dual-head SPECT in clinical practice.
Limitations of our study include the inevitable lack of a reference standard and not having an even larger patient population. However, by using a vigorous statistical analysis not previously applied in other reported investigations, we think that our study has convincingly shown that SPECT fails to show improvement over planar parallel-hole collimator imaging in detection, quantification, or localization of defects. Because of the longer acquisition time, the potential for motion artifacts, and the need for sedation in young children with SPECT, planar imaging may be preferable to dual-head SPECT in many imaging centers.
In conclusion, 99mTc-DMSA renal cortical imaging using dual-head SPECT offers no statistically significant diagnostic advantage over planar parallel-hole collimator imaging for the detection of focal defects.

Acknowledgments

For technical support, we thank Melvin Reinhardt, Chief Technologist, Supervisor, Division of Nuclear Medicine, Johns Hopkins University Hospital, Russell H. Morgan Department of Radiology. We also thank Ron Keren for adapting the artwork in Figure 1.

Footnote

Address correspondence to H. A. Ziessman ([email protected]).

References

1.
Eggli DF, Tulchinsky M. Scintigraphic evaluation of pediatric urinary tract infection. Semin Nucl Med 1993; 23:199 –218
2.
Maddahi J, Kiat H, Van Train KF, et al. Myocardial perfusion imaging with technetium-99m sestamibi SPECT in the evaluation of coronary artery disease. Am J Cardiol 1990; 66:55E–62E
3.
Fawcett HD, Sayle BA. SPECT versus planar liver scintigraphy: is SPECT worth it? J Nucl Med 1989; 30:57–59
4.
Gelfand MJ, Elgazzar AH, Kriss VM, et al. Iodine-12-MIBG SPECT versus planar imaging in children with neural crest tumors. J Nucl Med 1994; 35:1753 –1757
5.
Tarkington MA, Fildes RD, Levin K, Ziessman H, Harkness B, Gibbons MD. High resolution single photon emission computerized tomography (SPECT) 99mtechnetium-dimercapto-succinic acid renal imaging: a state of the art technique. J Urol 1990; 144:598–600
6.
Yen TC, Chen WP, Chang SL, Liu RS, Yeh SH, Lin CY. Technetium-99m-DMSA renal SPECT in diagnosing and monitoring pediatric acute pyelonephritis. J Nucl Med 1996; 37:1349 –1353
7.
Applegate KE, Connolly LP, Davis RT, et al. A prospective comparison of high-resolution planar, pinhole, and triple-detector SPECT for the detection of renal cortical defects. Clin Nucl Med 1997; 22:673 –678
8.
Mouratidis B, Ash JM, Gilday DL. Comparison of planar and SPECT 99Tcm-DMSA scintigraphy for the detection of renal cortical defects in children. Nucl Med Commun 1993; 14:82–86
9.
Everaert H, Flamen P, Franken PR, et al. 99mTc-DMSA renal scintigraphy for acute pyelonephritis in adults: planar and/or SPET imaging? Nucl Med Commun 1996; 17:884–889
10.
Piepsz A. Cortical scintigraphy and urinary tract infection in children. Nephrol Dial Transplant 2002; 17:560–562
11.
Majd M, Nussbaum Blask AR, Markle BM, et al. Acute pyelonephritis: comparison of diagnosis with 99mTc-DMSA, SPECT, spiral CT, MR imaging, and power Doppler US in an experimental pig model. Radiology 2001; 218:101–108
12.
Rushton HG, Majd M, Chandra R, Yim D. Evaluation of 99mtechnetium-dimercapto-succinic acid renal scans in experimental acute pyelonephritis in piglets. J Urol 1988; 5:1169 –1174
13.
Parkhouse HF, Godley ML, Cooper J, Risdon RA, Ransley PG. Renal imaging with 99mTc-labelled DMSA in the detection of acute pyelonephritis: an experimental study in the pig. Nucl Med Commun 1989; 1:63 –70
14.
Giblin JG, O'Connor KP, Fildes RD, et al. The diagnosis of acute pyelonephritis in the piglet using single photon emission computerized tomography dimercaptosuccinic acid scintigraphy: a pathological correlation. J Urol 1993; 150:759 –762
15.
Majd M, Rushton HG, Chandra R, Andrich MP, Tardif CP, Rashti F. Technetium-99m-DMSA renal cortical scintigraphy to detect experimental acute pyelonephritis in piglets: comparison of planar (pinhole) and SPECT imaging. J Nucl Med 1996; 37:1731 –1734
16.
Craig JC, Wheeler DM, Irwig L, et al. How accurate is dimercaptosuccinic acid scintigraphy for the diagnosis of acute pyelonephritis? A meta-analysis of experimental studies. J Nucl Med 2000; 41:986 –993
17.
Pediconi F, Catalano C, Occhiato R, et al. Breast lesion detection and characterization at contrast-enhanced MR mammography: gadobenate dimeglumine versus gadopentetate dimeglumine. Radiology. 2005; 237:45–56
18.
Altman DG. Practical statistics for medical research. London, UK: Chapman and Hall,1991

Information & Authors

Information

Published In

American Journal of Roentgenology
Pages: 333 - 337
PubMed: 19620428

History

Submitted: September 7, 2008
Accepted: December 8, 2008

Keywords

  1. 99mTc-DMSA
  2. acute pyelonephritis
  3. renal cortical scanning
  4. renal scintigraphy
  5. SPECT

Authors

Affiliations

Michele Brenner
Russell H. Morgan Department of Radiology, Division of Nuclear Medicine, Johns Hopkins University and Johns Hopkins Outpatient Center, 601 N Caroline St., JHOC Rm. 3231, Baltimore, MD 21287.
Dacian Bonta
Russell H. Morgan Department of Radiology, Division of Nuclear Medicine, Johns Hopkins University and Johns Hopkins Outpatient Center, 601 N Caroline St., JHOC Rm. 3231, Baltimore, MD 21287.
Hedieh Eslamy
Russell H. Morgan Department of Radiology, Division of Nuclear Medicine, Johns Hopkins University and Johns Hopkins Outpatient Center, 601 N Caroline St., JHOC Rm. 3231, Baltimore, MD 21287.
Present address: Department of Radiology and Molecular Imaging Program at Stanford, Division of Nuclear Medicine, Stanford University Medical Center, Stanford, CA.
Harvey A. Ziessman
Russell H. Morgan Department of Radiology, Division of Nuclear Medicine, Johns Hopkins University and Johns Hopkins Outpatient Center, 601 N Caroline St., JHOC Rm. 3231, Baltimore, MD 21287.

Metrics & Citations

Metrics

Citations

Export Citations

To download the citation to this article, select your reference manager software.

Articles citing this article

View Options

View options

PDF

View PDF

PDF Download

Download PDF

Media

Figures

Other

Tables

Share

Share

Copy the content Link

Share on social media