| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Duke University Medical Center, Box 3808, Durham, NC
27710.
2 Present address: Northside Radiology Associates, Atlanta, GA.
3 Present address: Department of Radiology, Montfort Hospital, Ottawa, ON,
Canada.
Received July 3, 2007;
accepted after revision August 15, 2007.
Supported by GE Healthcare.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. The study was HIPAA compliant and institutional review board approved. Fifty consecutive patients with suspected nephrolithiasis were recruited to undergo conventional renal stone unenhanced 16-MDCT with at least 160 mA. Noise was then artificially introduced to simulate levels of 70, 100, and 130 mA. Three blinded independent readers interpreted the original and simulated-dose scans for the location and number of renal and ureteral calculi and secondary signs of obstruction using a 5-point confidence scale.
RESULTS. Reader acceptability of scans was inversely related to noise. There was no significant reduction in readers' confidence in detection or exclusion of renal collecting system calculi with simulated reduction of mA of 70, 100, and 130 compared with the standard-dose study. However, for ureteral calcifications, there was a decrease in confidence for the detection or exclusion of ureterolithiasis at an mA of 70 (35 mAs).
CONCLUSION. An mA as low as 70 (35 mAs) is acceptable for evaluation of nephrolithiasis. However, the evaluation of ureterolithiasis is compromised with an mA of 70.
Keywords: helical CT • nephrolithiasis • radiation dose • ureterolithiasis
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
A limitation of unenhanced CT is that patients are exposed to a higher radiation dose compared with conventional radiography [7]. Further, many patients with stone disease require repeat imaging to assess for changes in stone burden. Previous studies with single-detector CT and 4-MDCT have shown that by reducing the tube current and thus the dose, diagnostic accuracy for detection of nephroureterolithiasis may be maintained [8, 9]. However, to our knowledge, it is not known what minimum dose is required for adequate stone detection using 16-MDCT.
Our purpose was to prospectively compare the reader acceptability and confidence of a range of reduced-dose 16-MDCT images with standard-dose 16-MDCT images for the detection of nephroureterolithiasis using a dose reduction simulation technique.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Scanning
All patients underwent CT with a 16-MDCT unit (LightSpeed 16, GE
Healthcare). Patients received no IV or oral contrast material. Patients were
scanned in the prone position. Scanning was performed from the superior aspect
of the kidneys to the inferior aspect of the pubic symphysis. In patients
weighing up to 200 lb (91 kg), the protocol was as follows: detector
configuration, 16 x 0.625 mm; 140 kVp; 160 mA; section thickness, 5 mm;
pitch, 1.75; table speed, 17.5 mm per rotatio; gantry speed, 0.5 second per
rotation. Tube current modulation was not used. Images were reconstructed
using a soft image algorithm. In patients in weighing more than 200 lb (91
kg), the mA was increased above 160 at the discretion of the technologist. For
purposes of analysis, all patients in whom the mA was increased above 160 were
analyzed on the basis of the maximum mA used.
Simulation Technique
The simulation technique and validation are described in greater detail
elsewhere [10]. The technology
uses a reduction model, which adds a random gaussian noise distribution to the
projection CT examination data. This technique adds noise based on the
selected tube current (mA) to simulate the amount of noise that would be found
if the examination were performed at this tube current. The simulation model
does not alter the mean signal level and adds noise on a sample-by-sample
basis (instead of one factor for the entire projection). Simulations were
performed on a dedicated research console that was offline but identical to
those used in the CT suite. Once image noise was added, an image data set
identical to that of the original examination was created (e.g., identical
start and stop levels, image thickness, and interval) except for the
annotation of the simulated tube current, which replaced the original tube
current annotation, in the left lower corner of each image. Simulated
examinations were listed under a new series number.
The original and simulated examinations were then sent to a research archive that was part of the existing clinical PACS (Centricity, GE Healthcare). Simulations represented serial tube current reductions to 160 (in patients in whom the original mA was > 160), 130, 100, and 70 mA. For this protocol the gantry rotation time was 0.5 second, so the mAs of the examinations were 80, 65, 50, and 35, respectively. Although the product of tube current and gantry cycle time (mAs) is a common descriptor for CT, we have chosen to discuss tube current alone because the simulations were based on changes in tube current only and all other parameters for the CT examination were identical.
Image Analysis
The original and simulated 5-mm axial images, all anonymized, were reviewed
in a random fashion on a PACS workstation used for clinical CT evaluation.
Three fellowship-trained abdominal radiologists with 1-14 years of experience
in abdominal CT interpreted the image sets independently. This wide range of
experience mimics our clinical practice. All annotations were removed from the
display, including the mA settings. Images were displayed using a standard 4:1
format at a standard soft-tissue window setting (width, 340 H; level, 40 H).
Readers were able to view images using any of the tools on the PACS as they
would when interpreting routine CT examinations. Room lighting was maintained
at a constant level for the period of review. Interpretations were completed
during a 6-week period.
For each image set, readers documented the number of stones identified in both the left and right renal (proximal) collecting system. In addition, readers were asked to assess specific signs of obstruction. The primary sign assessed was the presence of ureteral calcification. Secondary signs assessed included the rim sign, defined as a soft-tissue rim surrounding a calcification, pelvocaliectasis, perinephric stranding, periureteral stranding, asymmetry of renal size, and presence of a cyst or mass. Extrarenal findings were also assessed, including the presence of bowel wall thickening, mesenteric inflammation, free intra peritoneal fluid, enlarged lymph nodes, and pneumoperitoneum.
Readers assessed confidence of each finding on a 5-point scale as follows: 1 = definitely absent, 2 = probably absent, 3 = indeterminate, 4 = probably present, and 5 = definitely present. The readers' subjective impressions of the inter-pretative quality for each examination were also noted. Overall scan acceptability was assessed on a 5-point scale as follows: 1 = not acceptable, 2 = marginally acceptable, 3 = fairly acceptable, 4 = acceptable, and 5 = very acceptable. The readers' subjective impressions of the patients' body habitus were obtained. Body habitus was assessed on a 4-point scale as follows: 1 = thin, 2 = normal, 3 = moderately obese, and 4 = massively obese.
In addition, image noise (SD of Hounsfield units) was measured in each patient at three arbitrarily designated levels consisting of upper abdomen—the level of the mid kidney; mid abdomen—midway between the lower pole of the lowest kidney and the top of the iliac crests; and the pelvis—the level of the mid acetabulum. One of the authors performed all measurements. Variably sized, round regions of interest (ROIs) were placed on structures or organs (one ROI for each) at each level. ROI size was based on the organ or structure measured, and placement also avoided areas of artifact or inhomogeneity. Single ROIs were placed on the following organs or structures: upper abdomen—right kidney, left kidney, liver, right paraspinal muscle, aorta, and air adjacent to the anterior abdomen; mid abdomen—right paraspinal muscle, right psoas muscle, aorta, and air adjacent to the anterior abdominal wall; and pelvis—right gluteal muscle, right quadriceps group, bladder, and air adjacent to the anterior lower abdominal wall.
Statistical Analysis
For each qualitative variable, the mean response from the three readers was
compared at each tube current level using the paired signed rank test. A
p value of 0.01 or lower was considered significant. The difference
in image noise among the different simulated mA levels was assessed by means
of the repeated measures analysis of variance model (also called a mixed
model). The tube current level was used as the continuous variable. The
difference in image noise at the different simulated mA levels was broken into
categories on the basis of the body habitus rating of 1-2 or 3-4. The analysis
of variance model was used to estimate the added noise associated with a body
habitus score of > 3 versus < 3. Statistical software was used (SAS,
version 9.1, SAS Institute).
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
As expected, there was increased reader acceptability of scan quality with overall scan quality at higher simulated mA levels. At tube currents of 70, 100, 130, 160, and > 160 mA, overall reader acceptability (± SD) was 3.28 ± 0.6, 3.52 ± 0.64, 3.63 ± 0.68, 4.16 ± 0.55, and 4.28 ± 0.42, respectively. All comparisons achieved statistical significance except the comparison of 100 mA versus 130 mA and 160 mA versus the maximum mA. However, even at an mA as low as 70, a score of 3 or higher (on a 1-5 scale) was given to 70% (35/50) of cases, 92% (46/50) of cases, and 60% (30/50) of cases by readers 1, 2, and 3, respectively.
|
|
|
However, ureterolithiasis comparisons of 160 versus 70 mA, 160 versus 100 mA, and maximum versus 100 mA had a p value of less than 0.01 (Table 1). This indicates that at higher mA, significantly fewer cases are considered indeterminate for ureterolithiasis compared with lower mA.
A higher mA was generally associated with increased confidence for the presence or absence of both nephrolithiasis and ureterolithiasis (Table 2). For this assessment, variables are categorized as nephrolithiasis or ureterolithiasis definitely present or definitely absent. For this assessment, a score of definitely present (score of 5) was considered to be 1. All other scores (1, 2, 3, or 4) were considered to be 0. For the assessment of calculi definitely absent, a score of 1 was considered to be 1. All other scores (2, 3, 4, or 5) were considered to be 0. These variables were computed as reader averages for each mA level and the levels were compared with each other using the signed rank test.
|
Evaluation of the significance of this assessment shows that for nephrolithiasis definitely present or definitely absent there was a trend toward higher confidence with higher mA, but only the comparison of 70 versus 130 mA achieved statistical significance (Table 3). For ureterolithiasis definitely present or definitely absent, several comparisons achieved statistical significance (Fig. 2A, 2B, 2C, 2D, 2E).
|
|
|
|
|
|
Tube current had little effect on the confidence of identification of secondary signs associated with ureteral obstruction, including the presence or absence of rim sign, pelvocaliectasis, perinephric stranding, periureteral stranding, asymmetry of renal size, and the presence of a cyst or mass (Table 4). Of all the comparisons for the secondary signs of obstruction, only the 160 versus 130 mA for the presence of a rim sign proved to be statistically significant (p = 0.006).
|
The effect of a lower mA was discernible for the evaluation of extrarenal findings, including bowel thickening, mesenteric inflammation, enlarged lymph nodes, free intraperitoneal fluid, and free intraperitoneal gas (Table 5). For confidence in the determination of bowel thickening, mesenteric inflammation, and free intraperitoneal fluid, some of the comparisons with the lowest mA of 70 proved to be statistically significant. However, confidence in evaluation of enlarged lymph nodes and pneumoperitoneum was unaffected by changes in mA. Measurement of mean image noise in a variety of structures in the upper abdomen, mid abdomen, and pelvis indicates image noise increased inversely proportional to simulated mA, as expected with the simulation device (Tables 6, 7 and 8). The added noise (± standard error) in subjects with a body habitus score of 3 or 4 (moderately or massively obese) was 2.16 ± 1.2 in the upper abdomen, 3.17 ± 1.3 in the mid abdomen, and 2.41 ± 1.1 in the pelvis.
|
|
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Gottlieb et al. [4] have shown that the use of CT for suspected nephroureteral calculi has increased markedly since the introduction of unenhanced CT for urolithiasis in the mid 1990s. Relevant to the issue of radiation dose, the increase in use was most dramatic in women 30 years old or younger. Further, many patients with stones are at risk for recurrent ureteral colic and will likely need to undergo repeat imaging over time.
One major disadvantage of CT for the detection of urinary calculi is that the radiation dose is higher than conventional radiography [7]. Liu et al. [9] showed that even using a reduced-dose technique, the dose from single-detector CT is still estimated to be double that of a conventional excretory urogram. Other widely used single-detector CT protocols with thinner collimation and lower pitch have estimated radiation doses of 1.3-2.0 times higher [11-13] than the protocol reported by Liu et al. Recent work by Eikefjord et al. [14] found the average effective dose of 4-MDCT to be double that of excretory urography. Others suggest MDCT radiation doses may be higher than with single-detector CT [15].
Because most renal stones are radiodense and easily identified on unenhanced CT, reduced-dose techniques with increased image noise may prove to be sufficiently sensitive and specific for nephroureterolithiasis evaluation. In an anthropomorphic phantom with implanted stones in porcine kidneys, Spielmann et al. [16] reported excellent visualization of stones, even with significant reduction in the tube current (mA range, 170-20) and approximately 75% decrease in radiation dose. Indeed, many stones were visualized even at an mA as low as 20. In clinical work, reduced-dose scans obtained by decreasing mA or increasing the pitch have been shown to be both sensitive and specific for stone detection [8, 9]. Heneghan et al. [8] showed that in patients weighing less than 200 lb (91 kg), using both single-detector CT and 4-MDCT resulted in stone detectability similar to the standard-dose technique (reduction in the current from 160 to 76 mAs). In the study by Heneghan et al., dose reduction was 25% for 4-MDCT and 42% for single-detector CT.
The current study adds to this body of literature with a view toward determining reader confidence in detection of stones and reader acceptance of scan quality as reduced-dose techniques are used. The current work uses a novel CT simulation technique that essentially adds noise to reproduce the low-dose technique [10]. The advantages of this technology are that the scans with simulated reduced mA are identical to the original scans except for the addition of noise, and there is no need for additional radiation exposure to patients. This paradigm does not require radiation other than that from the original study, and multiple levels of simulated dose reduction, even to very low simulated mA, can be tested in a controlled fashion. In addition, a paradigm that simply lowers radiation dose empirically has the possibility of both false-negative studies (e.g., missing relatively subtle stones) and false-positive studies or decreased confidence for an interpretation of a normal study.
As expected, reader acceptability of scan quality was inversely related to noise. That is, acceptability levels were higher for the scans with higher mA compared with those with lower mA. This is to be expected because radiologists are accustomed to viewing scans relatively free of noise and likely find the noise-free scans more esthetically pleasing. It was interesting to note that even at a simulated mA as low as 70 (35 mAs), the readers found the majority of the cases to be fairly acceptable. However, this parameter of reader acceptability does not address stone detection per se.
For overall confidence in detection of either nephrolithiasis or ureterolithiasis, adequate confidence was achieved using a low-mA technique. There was no significant difference as a function of mA. However, there was a trend toward higher confidence with higher mA.
When confidence measures are viewed in terms of median deviation from the indeterminate assessment, similar findings are identified for nephrolithiasis. That is, mean deviation from the central assessment was similar regardless of mA. By contrast, significant differences were seen for detection of ureterolithiasis. That is, mean deviation from the central assessment was higher as a function of mA.
This study shows that when readers were asked whether nephrolithiasis was definitely present or definitely absent, there was no change in the assessment as a function of mA. This assessment differed when readers were asked whether ureterolithiasis was definitely present or definitely absent. For ureterolithiasis, a higher mA was associated with a definitely present or definitely absent assessment.
Our results suggest that even with an mA as low as 70 (an mAs of 35), several different reader parameters, including scan acceptability and confidence in detection for nephrolithiasis, are maintained. This finding suggests that relatively low-dose CT is comparable to a standard-dose CT for the detection and quantification of nephrolithiasis. The work corroborates the findings of Spielmann et al. [16], who showed in a phantom that with an mA as low as 20, detection of nephrolithiasis is maintained. The work also corroborates recent work by Poletti et al. [17] who found that even with a reduction of mAs down to 30, sensitivity for the detection of nephrolithiasis larger than 3 mm is maintained.
However, for ureterolithiasis, there was a dependency on mA. Specifically for scans with low dose, there was a significant difference in confidence in detection or exclusion of ureterolithiasis. Our result is in contrast to work by Poletti et al. [17], who found sensitivity and specificity of 4-MDCT for nephroureterolithiasis > 3 mm in size to be maintained even at 30 mA in patients with body mass index of less than 30. Similarly, Mulkens et al. [18] found that low-dose 4D MDCT using tube current modulation technique was accurate in the evaluation of patients with suspected renal colic. These contrasting results may be explained in part by calculi size. We did not stratify results by calculi size, but it is possible the calculi in our patient group were smaller than in other studies. Indeed, Tublin et al. [19] have shown that small calculi are more conspicuous at a higher mA setting. Similarly, Poletti et al. found that low-dose CT is less sensitive for calculi smaller than 3 mm compared with those larger than 3 mm. In addition, our study did not address sensitivity and specificity per se; rather, it assessed measures of reader confidence.
It can be estimated that the effective dose from our routine 16-MDCT renal stone protocol is 4.51 ± 0.45 mSv to 5.16 ± 0.70 mSv based on specific dose calculation methods [20]. With a reduction in dose from 160 to 70 or 100 mA, effective dose would be decreased by 56-38%. Another potential method to decrease dose would be to use a 16 x 1.25 mm detector configuration as opposed to the 16 x 0.625 mm configuration used in this project. We chose the 16 x 0.625 mm configuration because it is our routine for CT in the abdomen and pelvis for its benefits in image quality in multiplanar reformations. It should also be noted that our routine clinical renal stone protocol uses an mA of 160 (80 mAs), which is a lower-dose technique compared with routine abdominal and pelvic CT protocols [8, 10]. This relatively low-dose protocol is based on work from our institution where a reduced-dose technique was shown to be adequate for detection of nephroureterolithiasis. The current work suggests even further reductions in dose would be warranted.
There are limitations to this study that should be discussed. This project was limited to simulated reductions in mA using 16-MDCT images from one vendor; other factors that influence scan quality such as kVp, pitch, manufacturer, use of tube current modulation, and number of detector rows were not assessed. The evaluation was restricted to nephroureterolithiasis. Visualization of CT findings in clinical mimics of stone disease, such as acute appendicitis, was not addressed in detail. The lowest mA was 70 (mAs, 35), which did not allow determination of a lower threshold for confidence in detection of nephrolithiasis. Testing the images at an even lower mA may have identified such a threshold. There was a wide range of reader experience; however, the analysis reported mean reader scores, which may mitigate this effect. In addition, readers were aware that this project focused on dose reduction, which may have biased readers in favor of noisy images. Finally, the institutional review board approval was for comparisons of imaging studies. Clinical correlation of outcome measures were not performed; thus, there was no surgical or pathologic gold standard.
We conclude that for confidence in the MDCT evaluation of nephrolithiasis an mA as low as 70 (35 mAs) is acceptable for most cases. Such a low-dose technique may be particularly useful when following known nephrolithiasis. For confidence in the detection of ureterolithiasis, the evaluation is compromised at an mA as low as 70 (35 mAs). These mA values should be considered as 16-MDCT protocols are adopted.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. W. Ciaschini, E. M. Remer, M. E. Baker, M. Lieber, and B. R. Herts Urinary Calculi: Radiation Dose Reduction of 50% and 75% at CT--Effect on Sensitivity Radiology, April 1, 2009; 251(1): 105 - 111. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Karmazyn, D. P. Frush, K. E. Applegate, C. Maxfield, M. D. Cohen, and R. P. Jones CT with a Computer-Simulated Dose Reduction Technique for Detection of Pediatric Nephroureterolithiasis: Comparison of Standard and Reduced Radiation Doses Am. J. Roentgenol., January 1, 2009; 192(1): 143 - 149. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
