36 patients were scanned with Auto kV,
(mean age 55+/-19 (SD),
range: 16-81),
and size-matched to 36 control patients undergoing identical examinations (mean age 60+/-15 years,
range 17-89). There were 13 CT enterography exams,
13 CT urogram exams,
5 liver/pancreatic exams,
4 routine abdominopelvic exams,
and 1 CT angiogram. Lateral width in Auto kV patients was median 37.0 cm (range: 28 to 55),
compared to 36.5 cm in controls (range: 27-42). Of the 36 Auto kV exams,
the tube potential was changed to 80 kV in one exam,
100 kV in 30 exams,
and remained unchanged at 120 kV in 5 exams.
Dose Reduction Estimates
Mean ± (SD) CTDIvol was 16.02 ± 4.41(SD) mGy in the Auto kV exams.
The predicted CTDI if these patients had been scanned at 120 kV using the same base protocol was 19.54 ± 3.98 mGy,
a mean dose reduction of 3.52 ± 2.62 mGy,
or 18.5%. The mean CTDIvol in the matched control exams was similar at 19.34 ± 5.97 mGy,
which equates to a mean dose reduction of 3.32 ± 4.40 mGy,
or 17.2%,
in the Auto kV exams versus the matched controls.
The CTDIvol of Auto kV exams was significantly less than either the 120 kV prescription in the same patients or the CTDIvol of the control exams (p<0.001).
Dose reduction by patient size and kV
Figures 1 and 2 show the percentage dose reduction for the Auto kV patients,
with Figure 1 showing the change in dose between the 120 kV prescription and the Auto kV scan, according to patient size category. Dose reduction tended to be greater as patients measured smaller in lateral width (p=0.210),
and was greatest for the small-sized patients (median dose reduction of 26%).
The percent dose reduction between Auto kV cases and size-matched controls was similar (Figure 2). The greatest percentage dose reduction between the Auto kV cases and controls was achieved in the medium-sized patients,
where the median dose reduction was 24%.
Reduction in CTDIvol from Auto kV was also assessed by kV prescribed by the Auto kV tool (Figures 3 and 4).
For Auto kV patients,
there were significant changes in dose reduction according to the prescribed kV (p< 0.001),
with the large majority of Auto kV patients being scanned at 100 kV (30/36; 83%),
resulting in median percentage dose reductions of 28% (N=1 at 80 kV),
24% (N=30 at 100 kV),
and 2% (N=5 at 120 kV),
when compared to the estimated CTDIvol at 120 kVp in the same patient (Figure 3).
The dose reductions with Auto kV versus the size-matched controls are shown in Figure 4.
kV Selection and Image Quality
Two GI radiologists randomly assessed image quality in Auto kV and size-matched control CT exams,
blinded to kV and other parameters,
rating image sharpness,
noise,
noise texture and diagnostic confidence (Figures 5,6,7,8).
Sharpness,
noise,
noise texture and diagnostic confidence scores in the Auto kV and control exams for each reader are shown in the table (Figure 9 ). The Auto kV images were rated as very sharp (score= 1) in 33 (92%) and 36 cases (100%) by the two readers respectively,
versus in 36 (100%) and 35 (97%) of the control cases. Three Auto kV cases were rated as having questionable blurriness by only one reader,
as was one control case.
All 36 of the Auto kV cases and controls were scored by both readers as having optimal noise,
either similar to perceived routine noise (score 2a) or slightly more than routine but also adequate (score 2b) by both readers.
There was no noticeable change in noise texture in 32 (89%) and 33 (92%) Auto kV cases according to each reader. The readers had full confidence in 34 (94%) and 36 (100%) Auto kV cases; 1 reader scored “probably confident” (score= 2) in 2 cases (6%).
Iodine Contrast-to-noise Ratio (iCNR)
Figures 10-16 show box and whisker plots of noise,
iodine contrast and iCNR in cases and controls from the quantitative analysis. As expected,
the median attenuation was higher in liver (113 vs.
94 HU,
p =0.007) and portal vein (199 vs.
186 HU,
p =0.009) in the Auto kV cases compared to the control exams,
with the CT number difference in the aorta not reaching statistical significance (175 vs.160,
p =0.059). Similarly,
mean noise in the Auto kV cases was slightly and significantly higher in the cases than in the controls (19.4 HU vs 17.2 HU,
p = 0.007). In contrast,
however,
aortic iodine contrast-noise ratio (iCNR),
liver iCNR,
and portal vein iCNR were not statistically different (all p values>0.50) in the Auto kV cases compared to the controls.