CT Dose Index Protocol Optimization Case Study

This case study follows a clinical engineering and imaging physics review of an abdominal CT protocol. The scanner is producing diagnostically acceptable images, but the protocol dose metrics are consistently above the local diagnostic reference review level. The team must reduce unnecessary exposure without degrading the task-based image quality needed for the intended exam.

The case is useful because CT protocol optimization is not “use less dose” in isolation. It is a controlled engineering decision that connects radiation output, patient size, scan length, reconstruction setting, image noise, contrast-to-noise ratio, diagnostic task, uncertainty, workflow, and post-change quality assurance.

This is a simplified engineering example for protocol review. It is not a clinical recommendation, patient-specific dose estimate, or substitute for qualified medical physics, radiology, regulatory, or institutional review.

Case Context

A hospital reviews adult abdominal CT exams after a quarterly quality dashboard shows that one scanner’s routine protocol has higher dose metrics than comparable scanners. Radiologists report that the images are readable, but the clinical engineering team wants to know whether the exposure can be reduced while preserving the low-contrast detection task.

The engineering question is:

Can the protocol be released with lower tube-current settings and iterative reconstruction while keeping image quality above the defined acceptance threshold?

The answer depends on dose metrics, patient-size adjustment, dose-noise trade-off, CNR, and QA evidence after the change.

Simplified Protocol Data

Quantity	Symbol	Baseline value
volume CT dose index	CTDI_{vol}	18\ \text{mGy}
scan length	L	42\ \text{cm}
local review level for DLP		650\ \text{mGy cm}
patient effective diameter		30\ \text{cm}
simplified size conversion factor	f_{size}	1.25
baseline tube-current-time product		240\ \text{mAs}
proposed tube-current-time product		180\ \text{mAs}
baseline noise in water-equivalent region	\sigma_0	22\ \text{HU}
lesion-to-background contrast for task phantom	\Delta HU	80\ \text{HU}
minimum accepted task CNR		3.0
estimated noise reduction from reconstruction change		10\%
DLP uncertainty from scanner display and length record		\pm 8\%

The simplified terms used here are:

• CTDI_{vol}: scanner-reported normalized output metric for the protocol;
• DLP: dose-length product, a scan-length-weighted output metric;
• SSDE: size-specific dose estimate using a simplified size conversion factor;
• CNR: contrast-to-noise ratio for the defined imaging task.

These metrics are engineering controls and comparison tools. They do not fully describe organ dose, patient-specific risk, clinical appropriateness, or diagnostic performance by themselves.

Step 1: Baseline DLP

Dose-length product is:

With:

the baseline value is:

Compare with the local review level:

or:

Engineering Comment

The protocol is above the local review level. That does not automatically mean it is unsafe or inappropriate, but it does require engineering review. The next question is whether dose can be reduced without losing the diagnostic task.

Step 2: Baseline Size-Specific Dose Estimate

A simplified size-specific dose estimate is:

With:

the baseline estimate is:

Engineering Comment

The size factor increases the estimate because the patient is smaller than the reference phantom represented by the scanner output metric. SSDE is still a simplified estimate, but it is more informative than reading CTDI_{vol} alone without patient-size context.

Step 3: Proposed Tube-Current Reduction

The proposed tube-current-time ratio is:

For this screening calculation, assume CTDI_{vol} scales approximately with tube-current-time product:

New DLP:

New SSDE:

DLP reduction:

or:

Engineering Comment

The proposed setting brings DLP below the local review level and reduces the simplified SSDE. The calculation is only acceptable if automatic exposure control behavior, scanner limits, reconstruction version, patient-size range, and scan length remain controlled.

Step 4: Noise Increase from Lower Exposure

In a simplified x-ray quantum-noise screen, image noise scales approximately as:

With:

the predicted noise increase is:

Baseline noise:

Predicted noise before reconstruction improvement:

If the reconstruction change reduces noise by 10\%:

Engineering Comment

The new reconstruction setting nearly offsets the expected quantum-noise increase. That does not prove diagnostic equivalence. Reconstruction can change noise texture, edge behavior, low-contrast detectability, and measurement bias. The result is a screening calculation that must be checked on phantom and clinical review data.

Step 5: Contrast-to-Noise Ratio

Contrast-to-noise ratio for the task phantom is:

Baseline CNR:

Dose-reduced CNR before reconstruction improvement:

Dose-reduced CNR with reconstruction improvement:

Compare with the acceptance threshold:

Engineering Comment

The simplified CNR screen supports the proposed protocol, but only for the stated phantom contrast, reconstruction setting, and task threshold. If the clinical task requires subtler low-contrast detection, smaller patients, larger patients, metal artifact handling, or quantitative HU measurement, the threshold and validation data must change.

Step 6: Effective-Dose Screen for Review Communication

Some dashboards use a region-specific conversion factor for rough effective-dose communication. Suppose the review uses:

Baseline screen:

New protocol screen:

Reduction:

Engineering Comment

This rough effective-dose screen is useful for dashboard comparison, not for patient-specific risk prediction. Real patient dose depends on anatomy, scanner model, beam quality, modulation, organ location, patient size, shielding assumptions, scan range, and conversion method.

Step 7: Uncertainty Check

DLP uncertainty is estimated as:

High-side DLP after protocol change:

Compare with the local review level:

The high-side value remains below the review level.

For CNR, the team should also check uncertainty in phantom contrast, ROI placement, noise measurement, reconstruction version, and scanner calibration. If measured CNR after implementation falls near the threshold, the protocol should not be released on calculation alone.

Engineering Comment

Dose reduction is not enough. The protocol should pass with uncertainty included, and the image-quality metric should be measured under controlled conditions after implementation. A change that passes only in a spreadsheet is not release evidence.

Risk Review

The main hazards are:

Hazard	Failure mode	Engineering control
unnecessary radiation exposure	protocol uses more output than needed for the task	DLP/SSDE review, diagnostic reference comparison, protocol governance
missed finding	dose reduction increases noise or artifact enough to hide target	task phantom, CNR threshold, radiologist review, post-change audit
inconsistent protocol use	technologists select old or wrong protocol	protocol naming, access control, training, scanner preset review
hidden reconstruction change	software update changes noise texture or HU bias	version control, phantom comparison, QA acceptance record
patient-size mismatch	protocol works only for a narrow body-size range	size-stratified review and automatic exposure control checks

A simplified RPN screen for the missed-finding hazard before controls is:

After task phantom acceptance, protocol lockout, and post-change audit:

Engineering Comment

RPN does not prove safety. It helps prioritize controls and document why evidence is needed. The severity remains high because a diagnostic miss can matter clinically, so the team must rely on stronger controls than a lower calculated dose alone.

Decision

The proposed protocol can be released only after controlled QA evidence confirms the calculation. The engineering decision is:

1. approve the reduced tube-current protocol for phantom and limited supervised rollout;
2. require measured CNR at or above 3.0 with the new reconstruction version;
3. verify DLP and SSDE across representative patient-size bins;
4. lock protocol naming and scanner presets to prevent accidental old-protocol use;
5. document radiologist task acceptance for the intended abdominal indication;
6. monitor repeat-scan rate, image-quality complaints, and dose dashboard trends after release.

The change should not be released for every abdominal CT indication automatically. Different tasks may need different contrast, phase timing, motion tolerance, spatial resolution, or artifact handling.

Release Evidence

Evidence	Acceptance expectation
Scanner output	New CTDI_{vol} and DLP match expected values within tolerance.
Size adjustment	SSDE reviewed for representative patient-size bins.
Phantom image quality	CNR and noise texture meet the task threshold.
Reconstruction control	Software version and reconstruction setting are locked and documented.
Clinical review	Representative images accepted for the defined diagnostic task.
Protocol governance	Old preset removed or restricted; naming and training updated.
Risk record	Dose, missed-finding, workflow, and software-change hazards updated.
Post-release monitoring	Repeat exams, image-quality rejects, and dose dashboard trends reviewed.

Engineering Lessons

CT protocol optimization is a system decision. Dose metrics, image quality, reconstruction algorithms, patient size, workflow, and clinical task are coupled. Reducing CTDI_{vol} without checking CNR and task acceptance can harm diagnostic performance. Keeping a high-dose protocol without a justified task can expose patients unnecessarily.

The defensible engineering path is to make the trade-off explicit: define the task, calculate dose metrics, estimate the noise effect, validate image quality with phantoms and representative users, control the scanner configuration, and monitor the released protocol over time.