Halving the Dose With Pitch: Real or Illusory?
A technologist scanning a 70 kg patient for CT pulmonary angiography notices the protocol uses pitch 0.9, a fixed 300 mA at 120 kVp, and a 0.5 s rotation. To reduce dose, the technologist raises the pitch to 1.8 and changes nothing else. The console reports CTDIvol roughly halved and DLP roughly halved. The reconstructed images are noticeably noisier, and the radiologist remarks that the study "looks worse than usual." The technologist argues that the images were acceptable at the old settings, so the new protocol is a straightforward dose win. Scanner: 64-row MDCT, 40 mm detector width at the selected configuration.
1. Explain why CTDIvol fell by half while the noise increased, and confirm whether the DLP reasoning is correct.
Read the worked answer
Both numbers are right, and the images are correctly predicted to be noisier - these are not contradictory.
- For helical scanning, CTDIvol = CTDIw / pitch (pitch defined as table travel per rotation divided by beam width, using the same detector configuration). With everything else held constant, doubling pitch from 0.9 to 1.8 doubles the table speed and drops the dose per z-position by about half. CTDIvol approximately halves.
- DLP = CTDIvol × scan length. The anatomical coverage did not change, so DLP halves too. The arithmetic is correct.
- But the same table speed means the beam dwells half as long over each slice, so fewer photons per voxel. With mAs fixed, noise rises approximately with the square root of the change in photons: roughly a 41% increase in noise for a 2× reduction in photons. That matches the radiologist's impression.
The technologist's error is not in the arithmetic - it is in treating a lower dose number as a free improvement while the image quality changed. In CT, dose and image noise are linked; a dose drop at fixed mAs is paid for in noise.
2. Would the dose savings have survived if tube-current modulation with a noise index had been active? Explain what that tells you about where dose savings actually come from.
Read the worked answer
No - most or all of the savings would have evaporated.
- With a noise index as the target, TCM adjusts mA to hold image noise approximately constant. Doubling pitch removes photons per voxel; the modulator responds by raising mA to meet the same noise target.
- The product (mAs per rotation ÷ pitch) tends to converge back toward the value required by the noise index, so the CTDIvol the modulator settles at is close to the original. The dose reduction from pitch alone largely disappears.
- Conclusion: pitch is not a free dose lever. It is a speed lever. Real dose reduction at constant image quality comes from: stronger iterative/deep-learning reconstruction (accepting more noise in the raw data and removing it in the image domain), lowering kVp where contrast-to-noise allows, restricting z-coverage and eliminating overranging, avoiding unneeded phases, and correct patient centering.
The practical test before claiming a protocol improves dose: is the image quality demonstrably comparable? If yes, the savings are real. If the radiologist can see the difference, you traded image quality for dose, which may or may not be an acceptable trade for that clinical question.
3. For this CTPA specifically, what should have been changed instead - and what is the caveat about raising pitch that is not about dose at all?
Read the worked answer
Better levers for CTPA:
- Lower kVp to 100 (with the protocol validated by the physicist). Iodine attenuation rises sharply as kVp falls toward the iodine k-edge, so pulmonary arterial enhancement improves while dose drops; the cost is higher image noise, usually offset by iterative reconstruction. This is a standard CTPA dose-reduction move.
- Tighten z-coverage - scan apex to costophrenic angles, not the full chest out of habit - and let adaptive collimation reduce overranging at the ends of the helix.
- Stronger iterative or deep-learning reconstruction to absorb the noise.
- Confirm the 300 mA fixed technique is appropriate at all; modulation plus a noise index usually does better than a fixed mA across the thorax.
The non-dose caveat: raising pitch shortens the acquisition time proportionally. At pitch 1.8 with a 0.5 s rotation and 40 mm collimation, table speed is high enough that the scan may finish in around 2 seconds. If the bolus timing was set for a longer acquisition, or if the patient's breath-hold and the contrast transit do not line up, you risk scanning before or after peak pulmonary arterial enhancement - producing a study that is noisy and mistimed, which is worse than the original. Any pitch change in a bolus-timed study should be reviewed alongside the timing protocol, not just the dose display.