CT Dosimetry Playbook : Sandbox AI Approach

⚕️ EDUCATIONAL MEDICAL PHYSICS SANDBOX: For training in CT protocol optimization. Not for direct clinical application. Always consult your medical physicist and institutional policies.

Adult CT Dosimetry Playbook ⚛️

Physics, Centering, AEC, and AI-Ready Rules

Created by Dr. Sharad Maheshwari MD - imagingsimplified@gmail.com

A Comprehensive Teaching Module for Radiology Residents & Technologists

Navigation Menu

CT Physics Optimization 📊

Dose optimization is governed by the ALARA principle while preserving task-specific diagnostic sufficiency. Do not treat parameters as independent knobs, but as a coordinated operating point.

Master Control Model

Dose ≈ ƒ (Geometry, kV, mAs, Reconstruction, Scan Length)

• ▶ Geometry: Affects AEC scout estimation
• ▶ Tube Voltage (kV): Affects dose per photon & contrast
• ▶ Tube Current (mAs): Affects photon count (noise)
• ▶ Reconstruction: Affects noise tolerance

The True Control Variable 🎯

While we discuss mAs and kV, the actual control variable in modern CT is the Image Quality Target (Noise Index / Qref mAs). Without establishing a target noise tolerance, AEC systems cannot stabilize. Halving mAs increases noise by 41%.

Noise ∝ 1 / √mAs

Reconstruction Trade Function ⚡

Dose and reconstruction must be integrated. If Iterative Reconstruction (IR) strength increases, mAs can safely decrease. If Deep Learning (DL) is utilized, kV can decrease further without breaching the noise floor.

Dose Penalty ∝ kV² to kV³

🧮 Interactive Tool: Noise vs. Dose Simulator

Adjust mAs and kV to see their direct mathematical impact on radiation dose and image noise compared to a baseline 120kV / 100mAs scan.

Tube Current (mAs) 100 Tube Potential (kV) 120

Relative Dose 100%

Relative Noise 100%

Patient Centering & Geometry 🎯

Centering is not a technologist issue; it is a first-order physics variable. Miscentering drives 10 to 50 percent dose escalation through dual mechanisms.

🛑 Hard Rule: Scout Integrity

AEC decisions are entirely dependent on the scout image. If the patient moves after the scout, the entire dose plan is wrong. You must invalidate the AEC plan and reacquire the localizer.

1. The Bowtie Filter Mismatch

Bowtie filters attenuate peripheral rays to match typical cross-sections. Miscentering (vertical or lateral) pushes peripheral tissues into the unattenuated central beam or the highly attenuated periphery, causing massive surface dose spikes and noise compensation errors.

2. AEC & Lateral Miscentering

While vertical shift magnifies the scout image, lateral (horizontal) shift pushes anatomy into the thickest part of the bowtie filter. This causes unilateral photon starvation. The AEC detects this noise and wildly increases overall mAs to compensate.

• 1 cm shift = Up to 8% Organ Dose ↑
• 4 cm shift = Up to 35% Organ Dose ↑
• 6 cm shift = Up to 49% Surface Dose ↑

🧮 Interactive Tool: Miscentering Dose Penalty

Simulate the mathematical penalty of vertical and lateral table miscentering.

Vertical Shift from Isocenter 0 cm

Lateral (Horizontal) Shift 0 cm

AEC Scout Error +0%

Surface Dose +0%

CARE Engine Mechanics 🧠

Lowering kV alone causes mAs to rise. AEC behaves by compensating noise. Understanding this coupling is critical to protocoling.

Publication-Grade Synthesis

"The observed 40 percent dose reduction is attributable to correction of geometric miscentering, which normalized AEC-derived effective diameter and reduced mAs, combined with a reduction in tube voltage that decreased dose per photon. The interaction between CARE Dose4D and CARE kV under accurate centering conditions resulted in a coordinated reduction in photon quantity and energy while maintaining image quality."

🌊

CARE Dose4D (mAs)

Derives effective diameter from the scout and compares it to a reference phantom to maintain a constant noise target.

⚡

CARE kV Failure Modes

CARE kV is not universally beneficial. It fails when:

• Obese Patients: System hits mAs ceiling, causing severe noise.
• Metallic Hardware: Noise model breaks entirely.
• Non-Contrast Scans: No iodine benefit justifies the kV drop.

🧮 Interactive Tool: Automated AEC/kV Selector

See how a smart CDS algorithm adjusts scanning parameters based on patient size, identifying failure thresholds.

Patient BMI 25 kg/m²

Normal Weight

Optimized kV 100

Ref. mAs 150

Dose Savings 32%

Phase Selection Decision Tree ⏱️

Redundant multiphase scanning multiplies patient radiation dose. Use this diagnostic tree to determine the optimal, lowest-dose phase strategy based on specific pathology.

Primary Clinical Presentation -- Select Presentation -- Hematuria (Blood in Urine) Suspected Pancreatic Mass Incidental Liver Lesion High Speed Trauma Routine Abdominal Pain

Specific Suspected Pathology

🖱️

Select a clinical presentation to view the evidence-based protocol recommendation.

Recommendation

Required Phases:

Clinical Justification:

Dose Multiplier 1x

Master Dose Engine 🎛️

Simulate the complex interactions between physics variables, AEC thresholds, and tube limits simultaneously. Complete flexibility is enabled to allow simulation of contraindicated protocols.

Inputs & Anthropometrics

Patient BMI 25.0 kg/m²

Z-Axis Scan Length 40 cm

Vertical Miscentering 0 cm

Lateral Miscentering (Horizontal) 0 cm

Clinical Task Routine / General Low Contrast (Liver Mets) High Contrast (CTA)

Contrast Admin Iodine Enhanced Non-Contrast

Tube Potential (kV) 80 kV (Low) 100 kV (Optimized) 120 kV (Standard) 140 kV (Penetration)

Pitch Factor 0.6 (Slow) 1.0 (Standard) 1.4 (Fast)

Reconstruction FBP Iterative (IR) Deep Learning

Metal Hardware None Present

Phases 1 (Single Phase) 2 (Dual Phase) 3 (Triple Phase)

Volume CTDI (mGy) 0.0 AEC derived tube output

Total DLP (mGy × cm) 0.0 CTDI × Length × Phases

Simulated Noise 100% Relative to baseline

Effective Dose 0.0 mSv

SSDE Estimate 0.0 mGy

Req. Tube mA 0 mA

AI-Ready Decision Framework 🤖

A deployable, deterministic Clinical Decision Support (CDS) architecture with strict enforcement gates.

Deterministic Hard Rules

Rule 1: Centering Gate

IF |offset| > 2 cm:
  BLOCK scan
  REPOSITION patient
  REACQUIRE localizer

Rule 2: Scout Integrity

IF patient position changed AFTER scout:
  INVALIDATE AEC plan
  REPEAT scout

Rule 3: Physics-Derived mA Target

Attenuation ∝ e^{(μ × diameter)}
mAs ∝ Attenuation²
IF req_mA > 800: FLAG photon starvation

Rule 4: Reconstruction Compensation

IF IR/DL strength ↑ :
  ALLOW mAs ↓
ELSE:
  MAINTAIN baseline mAs

Failure Matrix & Response

Condition	Risk / Impact	Required Action
Off-center Patient	AEC misreads size; Dose ↑	Reposition & Re-scout
Obese (BMI > 35)	mAs hits ceiling; Severe Noise	Force High kV (120-140)
Low kV + FBP	Noise floor breached	Increase mAs or Enable IR
Metallic Hardware	CARE kV noise model breaks	Lock kV to 140; disable CARE kV
Non-contrast Scan	Unjustified noise increase	Lock kV ≥ 120

Technical Failure Modes ⚠️

Click cards to reveal the physics behind common adult CT errors.

Geometry Error

Patient Off-Centering

Positioning patient too close to the tube.

Physics Fact

Off-centering alters magnification on the scout image. AEC incorrectly assumes the patient is larger, drastically increasing tube current and dose.

Parameter Error

Using 80kV on BMI > 35

Attempting to lower dose/increase iodine contrast on obese patients.

Physics Fact

Photon Starvation. 80kV photons lack the energy to penetrate deep tissue. The system maxes out tube current (mA), but severe noise ruins the image.

Z-Axis Error

Overscanning

Including the neck in a Chest CT or lower pelvis in an upper abdomen scan.

Physics Fact

DLP (Dose Length Product) is directly proportional to scan length. Unnecessary coverage irradiates sensitive organs (thyroid/gonads) without diagnostic yield.

AEC Error

OBTCM Misuse

Using anterior dose reduction without proper centering.

Physics Fact

Organ-Based Modulation reduces anterior mA. However, if the patient is miscentered, the reduced-dose zone misses the target organs, negating dose savings.

CT Dose & Risk Assistant 🧮

An intelligent tool for patient-specific radiation dose and risk assessment.

Patient Information

Age at Exposure

Sex Female Male

Patient Diameters (cm)

Standard Measurement Locations

Measure on a single, representative axial CT image:

• Chest CT: At the level of the main carina.
• Abdomen/Pelvis CT: At the level of the liver/kidneys.
• Head CT: At the widest part of the skull.

Scan Parameters

Scan Region Head Neck Chest Abdomen-Pelvis Trunk (Chest, Abdomen, Pelvis)

CTDIvol (mGy)

DLP (mGy·cm)

Effective Diameter:

Patient Absorbed Dose (SSDE)

-

A size-corrected estimate of the dose absorbed by the patient (mGy).

Effective Dose (E)

-

A standardized measure of radiation risk for context (mSv).

Lifetime Attributable Risk (LAR)
[Model-Based Estimate]

-

Personalized cancer risk estimate from this scan (%).

Patient-Friendly Summary 📄

This report helps you understand the radiation from your recent CT scan. It is important to remember that your doctor has determined that the benefit of this scan for your health is greater than the very small risks associated with it.

Understanding the Dose:

• The radiation dose from this scan is estimated to be equivalent to the natural background radiation an average person receives over approximately - years.
• This is comparable to the radiation from about - standard chest X-rays.

Understanding the Risk:

Based on scientific models, the estimated additional chance of developing cancer over a lifetime from this single scan is about -. This is a very small increase to the average person's lifetime cancer risk.

Disclaimer: LAR calculations are highly model-dependent inferences based on BEIR VII guidelines. This is an estimate based on standardized epidemiological models and should be used for educational context, not individual clinical diagnosis.

Competency Assessment 📝

Test your knowledge on adult CT dosimetry, centering, and phase optimization.

Quiz Complete!

Your Score:

0 / 5

Guidelines & Literature 📚

1. ACR Appropriateness Criteria ®

Evidence-based guidelines assisting referring physicians and other providers in making the most appropriate imaging decisions for specific clinical conditions, particularly regarding phase reduction.

2. AAPM Report No. 204: Size-Specific Dose Estimates

Methodology for normalizing CT dose indices (CTDIvol) based on patient dimensions, providing a more accurate estimation of absorbed dose for varying adult body habitus.

3. RSNA: Impact of Patient Centering on Organ Dose

Demonstrates that vertical miscentering by up to 6 cm increases surface dose by 40 to 50 percent due to AEC scout magnification and bowtie filter mismatch. (Barreto et al.)

4. RSNA: Iterative & Deep-Learning Reconstruction

Literature confirms that replacing standard FBP with Iterative Reconstruction (IR) cuts dose by 30 to 60 percent, and Deep Learning Reconstruction (DLR) enables safe ultra-low kV protocols.

5. RSNA Quality Improvement Initiatives

Strategies focusing on standardizing tube potential selection, iterative reconstruction deployment, and eliminating non-indicated multiphasic protocols across health systems.

© 2026 Adult CT Protocol Optimization Playbook

Developed for medical physics educational demonstration based on established adult CT dosimetry principles and advanced RSNA literature.