PEDIATRIC ABDOMEN IMAGING: RESPONSIBILITY FIRST FRAMEWORK FOR RADIATION OPTIMIZATION (SANDBOX AI APPROACH)
⚠️ EDUCATIONAL SANDBOX ONLY: Not validated, not for clinical decision-making. All numbers are illustrative unless a source is shown. Verify against institutional, ACR and Image Gently guidance.
🚫 SYSTEM HARD STOP
CT not indicated as first-line 👶
Override accountability log (simulated)
In a real deployment, bypassing a hard stop would require a documented justification and be stored for QA review. In this sandbox nothing is stored.
PEDIATRIC ABDOMEN IMAGING
RESPONSIBILITY FIRST FRAMEWORK FOR RADIATION OPTIMIZATION (SANDBOX AI APPROACH) 🦋
Independent educational project. Not affiliated with or endorsed by the Image Gently Alliance, ACR, AAPM or ESR.
Radiation Biology & Risk 🧬
Why pediatric patients need specialized protocols, and why each extra phase adds exposure.
Why children are different 👶
✦ Higher cellular mitotic activity
✦ Greater organ radiosensitivity (e.g., thyroid, breast, marrow, gonads)
✦ Longer remaining lifespan for effects to appear. Risk projections rely on linear no-threshold assumptions, which remain debated.
Guideline synthesis
Image Gently principles
Child-size protocols. Prefer ultrasound/MRI where appropriate. Eliminate unnecessary phases.
ACR Appropriateness Criteria
Routine abdomen: a single portal venous phase is usually sufficient. Multiphase only for specific indications (e.g., active hemorrhage, selected tumors). Check the current ACR variant.
Illustrative effective dose (mSv) 📊
Illustrative values only. Replace with measured local data.
Indication & modality hierarchy 📋
Clinical indication
Preferred (1st line)
Optional
Rare (override)
Protocol decision engine ⚙️
Select an indication to simulate deterministic hard stops. Pillar scores are illustrative lookups, not validated measurements.
Real-time RATSe impact (illustrative)
Clinician override (simulated)
Not stored
Approved first-line pathway
CT phase rule
Contrast strategy
Dosimetry sandbox 🧸
Illustrative model: CTDIvol and benchmarks below are placeholders, not ACR Dose Index Registry data. Scan length changes DLP, not CTDIvol.
Target kV (auto)
IV contrast @ 550 mg I/kg
Venous delay (illustrative)
Est. abdomen length (height ≈ cm)
DRL benchmark check ⓘ
0 mGyBenchmark: mGy
Est. CTDIvol (illustrative)
SSDENeeds water-equivalent diameter (Dw) from the topogram. Use the Risk Calculator tab.
DLP = CTDIvol × length
Pediatric CT dose & risk illustration 🧮
A population-level teaching illustration of SSDE, effective dose and lifetime attributable risk (LAR). It is not an individual risk assessment.
Verify before use: the age-specific k-factors and the BEIR VII-derived LAR coefficients in this file were transcribed by hand and must be checked against AAPM Report 96 / Deak 2010 and BEIR VII before any teaching use. The SSDE table assumes the scanner reports CTDIvol for the 32 cm phantom and covers 15–42 cm (body) and 10–20 cm (head); values outside the range are not extrapolated.
SSDE (mGy)
Effective dose (mSv)
Illustrative LAR (%)
Effective dose applied to BEIR VII coefficients: an approximation.
Plain-language summary 📄
This estimate describes the radiation from one CT scan using population-level models. Your care team chose this scan for a reason, and you can ask them why it was needed and whether alternatives without radiation were considered.
🌍 About of average natural background radiation (assuming roughly 3 mSv per year).
🩻 Comparable to about chest X-rays (assuming roughly 0.1 mSv each).
Educational estimate only. Individual risk is uncertain and cannot be determined from this tool.
Priority failure modes 🎴
Click a card to reveal the rule.
Resident training module
Core competencies: indication-based selection, dose calculation with the tools above, and ethical imaging principles, ending in protocol-selection OSCEs.
Reporting & medico-legal
Defensible reporting documents the indication, the justified protocol, and the dose-optimization evidence. Example statement (adapt to local policy):
"Radiation optimized per pediatric protocol (Image Gently principles)."
Competency OSCE
Simulated cases across indications.
CASE 1 / 10SCORE: 0
Governance & medico-legal ⚖️
How the RATSe pillars map onto this decision-support sandbox.
RATSe pillar
Sandbox component
Reading the illustrative scores
🟢 1.0: adheres to the modeled pediatric pathway.
🟠 0.5–0.9: justified deviation (e.g., multiphase or salvage protocol accepted on clinical necessity).
🔴 below 0.5: weakly justified exposure or a hard-stop bypass without an independent clinical criterion.
These scores are hand-set lookups for teaching. They have not been validated for reliability or construct validity.
References & guidelines 📚
Key literature (verify each against PubMed/DOI before citing)
Design notes & limitations 🌟
Implementation & regulatory reality
Moving this logic into production would typically mean a SMART on FHIR clinical decision-support app inside the EHR. Software that automatically changes scanner protocols, or that does not meet the criteria for non-device clinical decision support (for example under the FDA's 2022 CDS guidance), would likely be regulated as a medical device. Requirements differ by jurisdiction and need local regulatory review. Prospective validation of dose, diagnostic yield and override behavior would be required before clinical use.
Comments
Post a Comment