44 cm Across: Photon Starvation in a 165 kg Patient

This is photon starvation with a beam-hardening contribution, not a reconstruction bug.

Mechanism:

• At 120 kVp, a 44 cm path of soft tissue attenuates the beam heavily; relatively few photons reach the detectors on the lateral projections through the flanks.
• Once the photon count drops low enough, electronic noise dominates the signal, and the reconstruction amplifies the inconsistency into streaks and a mottled "grainy" texture in the fattiest, least attenuating regions.
• Tube-current modulation cannot always rescue this. TCM will raise mA on the thick lateral projections, but it is capped by the generator/tube maximum and by the selected rotation time. If the maximum mA still cannot deliver enough photons through 44 cm, the modulation has done all it can and the image stays noise-limited.
• Positioning matters: an off-isocenter patient has a longer path on one side and can push the modulation to its limit earlier, and the localizer-based attenuation map is less accurate.

Key point: the liver looks adequate because the anteroposterior path is shorter. Noise is worst where the path is longest - the flanks.

Ranked roughly by benefit-to-cost in a patient this size:

1. Re-center the patient in the bore and re-scan the localizer. Cheapest fix, zero dose cost, occasionally enough on its own. Do this first.
2. Increase rotation time (0.5 → 0.8 or 1.0 s). Longer rotation allows a higher mA per rotation without exceeding tube limits, so the modulation can actually reach the photons it needs. Dose rises roughly with mAs, but this is often the only way to get above the noise floor in a very large patient.
3. Lower the noise index (e.g., 10 → 6–7). This permits more mA; dose rises. This is the explicit trade: you are buying image quality with dose.
4. Raise kVp to 140. More penetrating photons, less noise per mAs and less beam hardening - but lower iodine contrast and, at constant mAs, higher dose. With modulation the mA may fall, so net dose is often similar or modestly higher while noise drops.
5. Widen detector width / reconstruct thicker (0.625 → 1.25 mm acquisition, 3–5 mm for interpretation). More photons per voxel, less noise; no dose cost, but slight loss of z-resolution - acceptable for diverticulitis.
6. Stronger iterative or deep-learning reconstruction. Noise reduction without additional dose; confirm with the radiologist which kernel/strength the department validates.

Explicitly not useful: repeating the same protocol, or simply lowering pitch without changing noise targets.

Diagnostic reference levels are benchmarks for a standard-sized patient, not hard ceilings, and this is exactly the case where that matters.

Reasoning:

• CTDIvol is measured in a 32 cm acrylic phantom. In a 44 cm patient, the SSDE conversion factor is less than 1, meaning the patient's actual size-adjusted dose is lower than the displayed CTDIvol for the same technique - but the image is noisier for the same CTDIvol.
• So a displayed CTDIvol at or modestly above the reference level in a patient this size does not mean the same thing as it would in a 70 kg adult. The physics of attenuation is against you: equal dose buys less image quality.
• Conversely, the reference level exists to flag protocols that are habitually high, not to force a nondiagnostic scan. A nondiagnostic study that must be repeated delivers more dose than one adequate acquisition performed properly.
• The defensible path: optimize first (center, rotation time, noise index, reconstruction, slice thickness), document the justification for exceeding the reference level if you must, and route the decision through the radiologist and medical physicist rather than a technologist-vs-supervisor standoff at the console.

If this patient recurs in the practice, that is a protocol-review trigger for the physicist - a size-specific protocol is the correct long-term fix.