CT dose survey in adults: what sample size for what precision?

Area of Science:

• Medical Imaging
• Radiology
• Radiation Dosimetry

Background:

• Accurate radiation dose assessment is critical in computed tomography (CT).
• Variability in dose descriptors like CTDIvol and DLP can impact survey reliability.
• Current sample sizes may not adequately represent dose variability.

Purpose of the Study:

• To determine the variability of CTDIvol and DLP.
• To propose a minimum sample size for achieving precise dose estimations.
• To analyze dose variability based on body region and patient weight.

Main Methods:

• Collected CTDIvol and DLP data from 19,875 CT acquisitions across five body regions.
• Investigated variability with sample sizes from 10 to 1000 and different patient weight categories.
• Calculated 95% confidence intervals (CI95/med) for increasing sample sizes to determine required sample size for CI95/med ≤ 10%.

Main Results:

• Required sample sizes for CI95/med ≤ 10% varied from 15 to 900 depending on body region and dose descriptor.
• Small sample sizes (10-20 patients) resulted in mean CTDIvol and DLP values ranging from 0.50 to 2.00 times the actual value.
• Dose descriptor variability was high across all body regions and influenced by weight selection.

Conclusions:

• Sampling errors are substantial in CT dose surveys using small patient samples.
• Increasing sample size at least tenfold is recommended to reduce variability in CTDIvol and DLP.
• Accurate radiation dose monitoring requires larger, carefully selected patient cohorts.