Comparison of diagnostic accuracy of ultra low‐dose computed tomography and X‐ray of the kidneys, ureters and bladder for urolithiasis in the follow‐up setting

Urolithiasis is frequently followed up with a low‐dose computed tomography of the kidneys ureters and bladder (LD‐CTKUB) with doses typically less than 3 millisieverts. Although X‐ray is a lower dose (0.5–1.1 mSv) alternative for follow up, it has lower diagnostic accuracy and is limited to radiopaque calculi. This study aims to compare the diagnostic accuracy of sub‐millisievert ultra‐low dose CT (ULD‐CTKUB) against X‐ray KUB for the follow up of urolithiasis when both are compared against the standard of care of a low‐dose CT KUB (LD‐CTKUB).


Introduction
Urolithiasis is a common presentation to healthcare facilities in Australia.It affects approximately 1 in 10 Australians and has an annual incidence of 131 cases per 100,000 population. 1 The diagnosis and treatment of urolithiasis has seen increasing resource and financial costs with the rate of interventions doubling over the period of 1995-2010. 2maging with non-enhanced CT of the kidneys, ureters and bladder (CTKUB) is now established as the investigation of choice for stone burden in non-pregnant adult patients.CTKUB has excellent accuracy for detection of calculi, demonstrating a sensitivity of 95%-96% and specificity of 98%. 3 It can also assess for secondary signs of urinary tract obstruction including hydronephrosis, renal enlargement and perinephric fat stranding as well as identify alternative causes for pain. 3,4T, however, has the disadvantage of emitting ionising radiation.This is particularly relevant as renal calculi have a recurrence rate of 50%-70% at 10 years, [5][6][7] meaning that a considerable proportion of patients will be subjected to multiple follow-up scans in their lifetime.Radiation exposure therefore becomes a concern with 32% of patients undergoing CT for urolithiasis receiving an annual dose of >20 mSv in a registry of a single tertiary centre in the United States, exceeding the recommended annual dose to radiation workers as per the International Commission on Radiological Protection (ICRP). 8-ray of the kidneys, ureters and bladder (X-ray KUB) serves as an alternative to CTKUB for the follow-up urolithiasis and has the advantage of emitting lower doses of ionising radiation.The typical expected effective dose for an X-ray KUB is 0.5-1.1 mSv.9,10 Limitations for X-ray KUB include lower diagnostic accuracy than CTKUB, particularly with smaller stones and those located in the mid and distal ureter.11 X-ray KUB is also limited to following up radio-opaque stones that have been shown to be visible on plain radiographs at the time of initial CT diagnosis.11,12 Low-dose CTKUB (LD-CTKUB) is now commonly performed for the assessment and follow up of urolithiasis.A meta-analysis by Xiang et al. 13 showed that LD-CTKUB with an effective dose of <3 mSv had high sensitivity (93%) and specificity (97%).Such lower radiation scanning techniques can reduce the cumulative radiation exposure to patients undergoing follow-up CTs.
5][16][17][18] Recent advancements in CT technology such as automated exposure control and iterative reconstruction algorithms have allowed CTs to be performed at lower doses while mitigating the detriment to image quality and preserving diagnostic accuracy. 19,20We are not aware of any study that compares the diagnostic accuracy of ULD-CTKUB and X-ray KUB where the CT has a similar or lower radiation dose than X-ray.
This study aims to compare the diagnostic accuracy of a submillisievert ULD-CTKUB protocol with conventional X-ray KUB in the setting of urolithiasis follow up.Both are compared against a reference standard of a low-dose CTKUB (LD-CTKUB) which is the de facto gold standard in routine clinical practice.

Patient recruitment
This prospective study was approved by the Sydney Local Health District Ethics Committee.All patients referred to our Department for CT to follow up urolithiasis were considered for recruitment.The inclusion criteria were patient age ≥18 years and a known diagnosis of urolithiasis based on a prior LD-CTKUB available for assessment on our Department PACS.Patients were excluded if they were pregnant.They were also excluded if the follow up was for a calculus >10 mm as these are likely to be readily visible on all modalities and would not have helped in discriminating the diagnostic accuracy between ULD-CTKUB and Xray KUB.Informed consent was obtained from all the patients.All participants were given written information regarding the study protocol and reason for the extra radiation exposure.Biometric data including sex, age, height, weight and BMI of all patients was recorded.

Scanning protocol
All patients underwent the following scans: a LD-CTKUB acting as the reference for the presence of stones, ULD-CTKUB and X-ray KUB.All CTs were performed using a single helical scanner (General Electric Revolution 256 slices, Milwaukee, WI, USA).Following a scout topogram, scans were acquired in a craniocaudal direction in the prone position.The LD-CTKUB was performed with the following parameters: tube potential: 100 kV (dependent on the patient's weight); tube current: 150-350 mA (dependent on the patient's weight); collimation: 80 mm detector coverage, rotation time: 0.5 s, field of view: 36 cm, pitch: 0.992:1 and slice thickness: 0.625 mm.The ULD-CTKUB was performed with the following parameters: tube potential: 100 kV (dependent on the patient's weight), tube current: 40 mA (irrespective of patient weight), collimation: 80 mm detector coverage, rotation time: 0.5 s, field of view: 36 cm, pitch: 0.992:1 and slice thickness: 0.625 mm.
An adaptive statistical iterative reconstruction algorithm was applied to all CT acquisitions (True Fidelity, General Electric, Milwaukee, WI, USA).The CT Dose Index (CTDI) and Dose Length Product (DLP) were recorded for each CT scan.
X-ray was performed by acquiring a single anteriorposterior radiographic projection with the patient in the supine position.

Image interpretation
All reconstructed imaging data sets, including the most recent prior CT, were anonymised and transferred to the SyngoVia workstation (Siemens, Forchheim, Germany) for analysis.CT images were quantitatively assessed for noise by placing region of interest markers over the psoas major muscle and the abdominal wall subcutaneous fat at the level of L3.The attenuation standard deviation was recorded for both regions of interest in each CT.
The CTDIvol and DLP were recorded for each CT.The DLP included both the topogram and CT.Effective dose was calculated by multiplying the DLP by a tissue-weighting conversion factor of k = 0.015, specific for the abdomen and pelvis. 21ssessment of X-ray and CT for calculi was performed by three readers (two board-certified radiologists and one final year radiology registrar).The three readers each independently reviewed all ULD-CTKUB set of studies together in one session, and subsequently, all X-ray KUB set in another session with a minimum 2 week interval between them.They were also blinded to the reference LD-CTKUB findings while doing so.Per patient analysis was performed to assess for the presence of calculi (regardless of their number) in the kidneys, the ureters and the bladder.The presence of calculi in each of these three anatomical locations was recorded in a binary fashion as either 'yes' or 'no'.
This was followed by a per stone analysis where the number, size and location of each calculus was recorded for all three anatomical locations.Only calculi ≥2 mm were counted to a maximum of five calculi per kidney as calculi smaller than that are highly likely to pass on their own and do not require any clinical intervention.If >5 calculi were present in a kidney, only the five largest were assessed.

Statistical analysis
Statistical analysis was performed using IBM SPSS v20 Software (SPSS Inc, Chicago, IL, USA).The LD-CTKUB was used as the reference scan.Estimates of power were planned based on recruitment sizes of prior studies and calculating power around per-patient rather than perstone due to the poor predictability of calculi per patient.
Image noise and dosage were compared between LD-CTKUB and ULD-CTKUB using the student t-test for intramuscular and fat attenuation standard deviation, DLP and effective dose.The Mann-Whitney U test was used for the comparison of CTDI.
For the per-patient analysis, the sensitivity and specificity were calculated for each.Sensitivity and specificity were compared between ULD-CTKUB and X-ray using the McNemar test.The per-patient sensitivity and specificity of ULD-CTKUB was also compared between overweight/obese patients and those with a BMI <25.
For the per-stone analysis, each calculus was recorded as either a successful detection or missed detection based on comparison to the reference LD-CTKUB.The detection rate was compared between ULD-CTKUB and X-ray using the McNemar test.
Interobserver agreement was assessed by performing intraclass correlation (average, two-way mixed).

Patient cohort
Between February 2018 and July 2021, a total of 68 patients were enrolled in the study.There were 42 males and 26 females.The patients ranged from 27 to 85 years with a mean age of 61.7 years and median of 63.3 years.The mean BMI was 28.5 kg/m 2 , including 18 overweight patients (BMI 25.0-29.9kg/m 2 ) (26.5%) and 27 obese patients (BMI ≥30.0 kg/m 2 ) (39.7%).

Diagnostic performance
A total of 197 calculi were identified on reference LD-CTKUB (Table 1).This included 176 in the kidneys (89.3%), 16 in the ureters (8.1%) and five in the bladder (2.5%).Of the 197 calculi, 111 measured <5 mm (56.3%) and 86 measured ≥5 mm (43.7%).No patients were found to have more than one ureteric calculus or more than one bladder calculus.Figures 1 and 2 outline examples of images seen on LD-CTKUB, ULD-CTKUB and X-ray.
The pooled sensitivities of ULD-CTKUB and X-ray for each location are summarised in Table 2.

Per-stone analysis
For all calculi regardless of location, ULD-CTKUB had a pooled detection rate of 79% and X-ray 48% (P < 0.01) (Table 4).ULD-CTKUB had a significantly higher detection rate than X-ray across all three readers and across both calculi size groups of <5 mm and ≥5 mm (P < 0.01) (Table 5).
For intrarenal calculi, ULD-CTKUB had a pooled detection rate of 81% and X-ray 50% (P < 0.01) (Table 4).ULD-CTKUB had a significantly higher detection rate than X-ray across all three readers and across both calculi size groups of <5 mm and ≥5 mm (P < 0.01) (Table 5).

Interobserver agreement
The overall interobserver agreement with regard to the per-patient analysis ranged from k = 0.72 to 0.94.For the detection of all calculi, there was strong agreement with k = 0.91 for ULD-CTKUB and k = 0.86 for X-ray.For the detection of intrarenal calculi, there was strong agreement with k = 0.94 for ULD-CTKUB and k = 0.91 for X-ray.For the detection of ureteric calculi, there was moderate agreement with k = 0.78 for ULD-CTKUB and k = 0.72 for X-ray.

Impact of BMI
In a subgroup analysis of patients who were overweight or obese (BMI > 25) versus not obese/overweight (BMI ≤ 25), ULD-CTKUB had lower pooled sensitivity in obese patients when compared to LD-CTKUB (84% vs. 98%, P < 0.01).There was no significant difference in pooled sensitivity for the detection of ureteric or bladder calculi.There was no significant difference in pooled specificity for the per-patient detection of calculi, regardless of location.

Discussion
Our ULD-CTKUB protocol showed high sensitivity (93%) and reasonable specificity (90%) for the diagnosis of urolithiasis on a per-patient basis, and this was significantly higher than that of the sensitivity of X-ray (67%) when compared to a reference LD-CTKUB.ULD-CTKUB also showed significantly higher sensitivity than X-ray for the assessment of intrarenal calculi on a per-stone basis.We demonstrated a strong interobserver reliability for ULD-CTKUB in the overall diagnosis of urolithiasis as well as for diagnosis of intrarenal calculi.
Urolithiasis is a frequently encountered condition in both inpatient and outpatient settings.Ongoing follow up forms an important component of its management with CT imaging being the established, gold-standard approach. 12,22The long-term risks of high radiation exposure become a concern in those patients undergoing repeated CTs for recurrent urolithiasis, particularly with  3. Individual reader per-patient sensitivity and specificity for ULD-CTKUB and X-ray.n = denominator for sensitivity and specificity regard to the development of secondary malignancy. 8,23urthermore, CT examinations are being performed more frequently as imaging services become increasingly accessible. 246][27] Further to this, CT with submillisievert dosage has also been achieved by a number of studies. 14,15,17,18The ULD-CTKUB protocol in our study achieved a mean effective dose of 0.5 mSv; a 75% dose reduction from our LD-CTKUB.This is lower than that of an X-ray KUB which does not have a negligible radiation burden with an effective dose reportedly ranging from 0.5 to 1.1 mSv. 9,10X-ray for urolithiasis has also been shown to have an effective dose that is similar to or in some cases higher than LD-CTKUB, particularly among obese patients. 28,29ur study did however show a lower per-stone sensitivity (79%) than other studies which have reported sensitivities of greater than 90% with LD-CTKUB. 14,26,30While the sensitivity for calculi ≥5 mm was comparable to other studies at 92%, calculi <5 mm were detected with a sensitivity of 68%.This decrease in sensitivity with smaller stones has been established in prior studies. 16,17he majority of calculi detected in this study were <5 mm.As per the European Association of Urology and American Urological Association, a clinically significant stone is defined as being ≥5 mm. 31,32Stones of this size are less likely to pass spontaneously and are more likely to require definitive treatment, usually operative management. 32Our study adopted a more conservative cut off of ≥2 mm.This allowed for the inclusion of stones 2-5 mm that are more likely to be the target of follow up with ULD-CTKUB.Our study importantly showed that ULD-CTKUB had a higher detection rate than X-ray for these smaller calculi <5 mm.
In addition to being more likely to undergo definitive treatment, calculi ≥5 mm are also more likely to be detected on X-ray at the time of initial diagnosis. 33Nevertheless, our ULD-CTKUB protocol was also found to have a higher detection rate than X-ray for calculi ≥5 mm.Traditionally, the lower diagnostic accuracy of Xray has been broadly viewed as an acceptable exchange for lower radiation dose. 28This paradigm is challenged with CT protocols of equivalent or lower dose than X-ray.
We report a higher sensitivity for the detection of ureteric and bladder calculi with ULD-CTKUB compared to Xray with reference to the LD-CTKUB gold standard.This reached statistical significance after the results of our three readers were pooled.On an individual basis, only one reader had significantly higher sensitivity for ureteric calculi while none of the readers reached significance for bladder calculi.The results of our individual readers are most attributable to the relatively low numbers of ureteric and bladder calculi in our cohort.This result is of clinical importance as obstructive ureteric calculi constitute the most common presentation of urolithiasis to emergency settings, often forming the basis for ongoing follow up.The detection of bladder calculi is also useful as it indicates those which have successfully passed through the vesicoureteric junction and no longer require monitoring or treatment. 12evertheless, we report a relatively lower sensitivity for the ULD-CTKUB detection of ureteric and bladder calculi compared to intrarenal calculi.This may be a result of a lower CT dose resulting in a particularly marked decreased image quality in the pelvis, thereby affecting the detection of calculi in the distal ureters and bladder.Of relevance to this, ULD-CTKUB was also found to have significantly higher quantitative noise than the reference CT.
The main strengths of our study include the consistent use of a reference LD-CTKUB and the prospective recruitment of study participants.The incorporation of a prior CTKUB for comparison purposes during image reading allowed the study to more closely replicate the process of urolithiasis follow up and thereby offer greater applicability to everyday practice.
Our study is limited to the follow-up setting, resulting in a higher pre-test probability of detecting calculi.It is expected that our patient cohort would have a lower number of patients without urolithiasis compared to a cohort with no known diagnosis.This likely enabled us to identify a larger number of calculi and thereby sufficiently power our study.The number of patients who received additional radiation as part of this study but had no urolithiasis would also likely be lower.This study did not account for patients with alternate diagnoses that could be missed because of the low dose protocol.These would however be unlikely to occur in the follow up setting.Furthermore, the use of LD-CTKUB as the reference standard has inherent limitations as it does not detect every calculus with 100% reliability.It would have been ideal to also include follow-up clinical data such as witnessed stone passage, disappearance of calculus on subsequent CT or direct visualisation on cystoscopy/ ureteroscopy.However follow up occurs over a highly variable time frame and is frequently in the private setting external to our institution.
We have successfully achieved a very low dose with our ULD-CTKUB protocol that is less than that of X-ray KUB.Our study has also compared the diagnostic accuracy of ULD-CTKUB with X-ray.Our results suggest that ULD-CTKUB can be used in place of X-ray for the follow up intrarenal calculi with higher sensitivity and lower dose.Despite CT being known to have greater diagnostic accuracy for urolithiasis, its higher dose has impeded its use in follow up, especially in patients of younger age or those who require recurrent imaging or long-term monitoring. 34The results of our study indicate that this can be somewhat ameliorated while maintaining high sensitivity and specificity.
The relatively low number of bladder and ureteric calculi is one of the weaknesses of this study.There is an additional radiation dose which results from prospective studies such as ours.This makes the inclusion of larger numbers of patients challenging from an ethical and informed consent perspective.Our study did however find a significantly higher pooled sensitivity for the detection of bladder and ureteric calculi with ULD-CTKUB, suggesting its potential role for the follow up of these calculi.Imaging in particular plays a key role in the follow-up ureteric calculi <5 mm. 12 While our study's results can be attributed to the follow-up setting, they are not applicable to diagnostic purposes such as the assessment of renal colic and haematuria.This is a trade off that occurs with a submillisievert protocol in its current state.This is evident in our study with the ULD-CTKUB protocol having a lower diagnostic accuracy than the LD-CTKUB.It is therefore imperative that medical professionals involved in the investigation and management of urolithiasis, including radiologists and clinicians, understand the current role and limitations of ULD-CTKUB.
In conclusion, this study shows that ULD-CTKUB has surpassed X-ray in diagnostic accuracy at lower doses, for the follow up of intrarenal calculi.Our pooled analysis across three readers also shows higher sensitivity than X-ray for the follow up of ureteric and bladder calculi.The results of our study hold particular significance for patients undergoing long-term follow up for urolithiasis who may be able to undergo CTs with a lower cumulative radiation exposure.The wider adoption of low-dose protocols will be reliant upon whether radiologists and clinicians are able to accept the resultant reduction in sensitivity.

Fig. 1 .
Fig. 1.Examples of LD-CTKUB (images a, d and g), ULD-CTKUB (b, e and h) and X-ray (c, f and i) in three patients.Calculi have been labelled with white arrows.Image (a) shows a 3 mm calculus in the left kidney, also visible on ULD-CTKUB (Image b) and X-ray (Image c).Image (d) shows a 2 mm calculus in the left kidney, not visible on either ULD-CTKUB (Image e) or X-ray (Image f).Image G shows a 5 mm calculus in the left kidney, visible on ULD-CTKUB (Image h) but not identified on X-ray (Image i) due to overlying bowel gas.

Fig. 2 .
Fig. 2. Examples of LD-CTKUB (images a, d and g), ULD-CTKUB (b, e and h) and X-ray (c, f and i) in three patients.Calculi have been labelled with white arrows.Image (a) shows a 6 mm calculus in the right ureter, also visible on ULD-CTKUB (Image b) but not on X-ray (Image c).Image (d) shows a 4 mm calculus in the left ureter, not visible on either ULD-CTKUB (Image e) or X-ray (Image f).Image (g) shows a 6 mm calculus in the bladder, also visible on ULD-CTKUB (image h) but not confidently seen on X-ray (Image i) due to the presence of multiple phleboliths.

Table 1 .
Number of calculi detected in the patient cohort by the reference LD-CTKUB, organised by anatomical location and size © 2023 The Authors.Journal of Medical Imaging and Radiation Oncology published by John Wiley & Sons Australia, Ltd on behalf of Royal Australian and New Zealand College of Radiologists.

Table 2 .
Pooled per-patient sensitivity and specificity for ULD-CTKUB and X-ray when compared to reference LD-CTKUB

Table 4 .
Pooled detection rate of calculi on a per-stone basis, for calculi in all locations and calculi in the kidney, divided by size group

Table 5 .
Individual reader detection rate of calculi on a per-stone basis, for calculi in all locations and calculi in the kidney, divided by size group.
© 2023 The Authors.Journal of Medical Imaging and Radiation Oncology published by John Wiley & Sons Australia, Ltd on behalf of Royal Australian and New Zealand College of Radiologists.