CT is the best and the first line imaging examination for urolithiasis.
As increased fear of the potential cancer risk of radiation exposure several studies to reduce the effective dose in urolithiasis have been published,
and they reported an effective dose ranging from 0.7 to 2.8 mSv [17,
18].
Our study showed an average of 1.48 mSv for an effective dose in the LD protocol,
similar to previous reports.
Despite the low effective dose,
a previous meta-analysis study of low-dose urolithiasis CT showed high sensitivity and specificity [18].
The effective dose of intravenous pyelography is reported to range between 0.7 to 3.7 mSv,
but the diagnostic performance was significantly lower than CT [10,
19].
Thus,
LD protocols have strengths including a higher diagnostic performance and no need for contrast material in similar effective dose.
For the ULD protocols,
the assumed effective dose was 0.64 mSv and it was very low,
even comparable to conventional single plain radiography of the kidneys,
ureters and bladder [19].
During the diagnosis and treatment of urolithiasis,
the average of four imaging studies including 1.7 CT is performed in the year after initial event [20].
Moreover,
there have been increasing trends showing a 1.4 % increase in incidence and a relatively high recurrence rate for urolithiasis,
so increased cumulative doses can raise the risk of cancer in the patient with urolithiasis [3,
7,
17].
Thus,
LD or ULD protocols can potentially contribute to reduce lifetime cancer risk.
The beam energy is determined by the peak kilovoltage of the tube and has an exponential relationship to the effective radiation dose.
The kilovoltage of the tube has a complex effect on image noise and tissue contrast,
so maintaining constant image quality is one of the purposes in radiation dose reduction [14].
Several ways to obtain better images have been researched and applied in current clinical studies.
In these manners,
lowering the peak kilovoltage,
increasing the milliampere-second,
and using automated tube current modulation are regarded as the best strategies for radiation dose reduction and several studies have reported [14,
18].
Furthermore,
depending on the development of model-based IR algorithms,
several investigators have published reports using the algorithms and they found it to be useful in image quality with more effective dose reduction [15,
21-24].
However,
it is more time consuming during the reconstruction phase,
taking between 10 and 90 minutes depending on the number of slices,
making model-based IR to inappropriate for urgent examinations [25,
26].
Nevertheless,
new computer technique has reached a tolerable rate of speed using knowledge-based IR [27].
The image reconstructed with knowledge or model-based IR shows an improvement in spatial resolution and image noise especially in low-dose or ultralow-dose scans,
but the scans appear differently,
often described as “waxy” or “plastic” as compared with FBP [26].
In our study,
IMR represents the best objective image noise,
despite of ULD protocols like other previous reports.
Though IMR did not show a comparable assessment to the RD or LD protocols,
statistically significant improvements were seen compared to FBP and iDose using ULD protocol and the subjective assessment scores were still higher.
Based on this,
it can provide better image quality in terms of subjective image quality,
noise,
and diagnostic confidence with proper dose modulation.
In the diagnostic performance of LD CT for urolithiasis,
our study showed overall concordant rates of 88.2% and 84.3% (staff radiologist and radiology resident,
respectively) compared to standard reference.
Although relatively lower concordant rates of 79.6% and 73.5% in case of renal stones < 3mm,
additional diagnostic or therapeutic procedures may not be required clinically because stones < 5mm pass spontaneously up to 98% of the time [1].
For the ureter stones,
the overall concordant rates were 97.8% and 91.1%,
but in cases of stones ≥ 3mm,
both the staff radiologist and the radiology resident achieved a concordant rate of 100%.
In cases of ureter stones < 3mm,
the concordant rate of the radiology resident was substantial of 63.6%.
However,
because of the lower clinical significance in stones < 5mm,
LD image can provide an image strong enough for good diagnostic performance even in less-experienced interpreters [1].
There were several limitations in our study.
First,
additional radiation was required for the study,
which can be an ethical problem.
However,
the ULD protocol scan was only performed over the a limited range,
and the actual total effective radiation dose of each patient (ranged 4.3 to 15.9 mSv) did not exceed the range of regular dose CT protocols previously reported [18].
Second,
a reference standard was not confirmed with the extraction or excretion of stones.
In particular,
it is debatable whether subtle and small stones in RD are true stones or not.
However,
we tried to reduce this limitation by using a consensus panel.
Third,
personal variation,
such as BMI,
which can vary the radiation dose or image quality,
was not considered in this study.
Moreover,
the relatively small number of patients involved in this study within one institution may not be sufficient to show statistical differences.
Although the reviewer was blinded,
experienced staff radiologists may be able to distinguish between 5 image sets due to the difference in image texture.
Thus,
bias may be present in the subjective image assessment.
Finally,
LD and ULD acquisitions were performed only in cases of positive initial CT.
Thus,
reviewers that interpreted the LD series were also biased.
However,
we believe that our study offers a number of positive features.
The different CT protocols were performed at the same time to allow for the exact comparison of virtually identical stones,
whereas most of the previous studies had time interval.
Furthermore,
our study compared RD using FBP reconstruction and LD using IR and evaluated diagnostic performance.
There can be a corresponding diagnostic performance of ULD using IMR because of the lowest objective image noise.
In conclusion,
knowledge-based IR can provide better objective image noise and subjective image assessment of a condition using the same radiation dose as other reconstruction algorithms.
Furthermore,
LD protocols showed a comparable diagnostic performance to RD protocols.
Although knowledge-based IR didn’t show better image quality compared to RD or LD protocols,
its potential strength can be expected in LD protocols.
To decrease effective radiation dose,
further studies are needed to confirm the diagnostic performance of ULD protocols using knowledge-based IR.