Clinical Scorecard: Automated Precision in Patient Positioning for Computed Tomography Utilizing Anterior-Posterior Localizer Images and a Deep Learning Approach: Findings from a Dual-Center Investigation

At a Glance

Key Highlights

• High prevalence of patient mis-centering in CT scans leads to increased radiation dose and degraded image quality.
• Deep learning using AP localizer images can automate detection of patient body centerline relative to gantry isocenter.
• Dual-center study with 5754 chest CT cases demonstrated feasibility of deep learning for automatic patient positioning.

Guideline-Based Recommendations

Diagnosis

• Evaluate patient centering using anterior-posterior localizer images prior to CT scanning.

Management

• Implement automated patient positioning systems using deep learning algorithms to reduce mis-centering errors.
• Avoid reliance solely on 3D cameras due to calibration complexity and interference from external objects.

Monitoring & Follow-up

• Continuously assess patient positioning accuracy and radiation dose metrics to ensure optimal scan quality and safety.

Risks

• Mis-centering can increase surface and organ radiation doses by up to 100%, degrade image quality, and affect automatic exposure control systems.

Patient & Prescribing Data

Chest CT patients from two imaging centers using Siemens Somatom Duo and Phillips Brilliance 16 scanners

Deep learning models trained separately per scanner type improved patient centering accuracy using only AP localizer images.

Clinical Best Practices

• Use deep learning-based automatic positioning to minimize patient mis-centering in chest CT scans.
• Exclude cases with truncation artifacts from automated positioning workflows to ensure accurate centerline detection.
• Train scanner-specific models due to differences in localizer image characteristics across CT systems.

References

• Akintayo et al - Prevalence of patient mis-centering in CT
• Sukupova et al - Clinical mis-centering prevalence
• Li et al - Surface dose increment due to mis-centering
• Furukawa et al - Effect of table height on tube current modulation
• Euler et al - Organ dose changes due to mis-centering
• Kaasalainen et al - Organ dose and CTDIvol changes with table height
• Booij et al - Automatic positioning using 3D camera
• Dane et al - Positioning error using 3D camera and AP localizer
• Gang et al - Impact of positioning correction on image quality and dose
• Saltybaeva et al - Reduction of manual positioning errors with 3D camera
• Deep learning applications in CT imaging