Estimating organ doses from tube current modulated CT examinations using a generalized linear model.

Area of Science:

• Medical Physics
• Radiological Dosimetry
• Computational Imaging

Background:

• Current Computed Tomography (CT) dose metrics often rely on fixed tube current simulations or physical measurements, failing to account for Tube Current Modulation (TCM).
• Existing dose metrics like size specific dose estimate (SSDE) do not accurately reflect actual patient organ dose, particularly with variable scan parameters.
• There is a need for advanced CT dosimetry models that can incorporate patient-specific factors and dynamic scanning techniques for precise dose estimation.

Purpose of the Study:

• To develop and validate a novel dose estimation model using the Generalized Linear Model (GLM) for CT examinations.
• To enhance organ dose estimation accuracy by incorporating Tube Current Modulation (TCM) data and regional patient size descriptors.
• To compare the performance of the GLM-based model against established dose metrics such as SSDE and ImPACT.

Main Methods:

• Collected 332 patient CT scans (pediatric/adult, thoracic/abdomen/pelvis) across three scanner systems, obtaining detailed TCM data where available.
• Performed manual organ segmentation and used Monte Carlo (MC) simulations to establish reference organ doses (lungs, breasts, liver, spleen, kidneys).
• Trained and evaluated GLM-based dose estimation models using approximately 60% and 40% of the data, respectively, exploring models with and without categorical variables.

Main Results:

• The Generalized Linear Model (GLM) provided organ dose estimates significantly closer to Monte Carlo (MC) reference values compared to SSDE and ImPACT.
• GLM performance surpassed SSDE and ImPACT regardless of the inclusion of categorical variables (e.g., scanner model, organ type).
• Inclusion of categorical variables notably improved breast dose estimation, with minor improvements for other organs.

Conclusions:

• The GLM approach offers a significant advancement in CT dose estimation, accurately assessing organ dose from Tube Current Modulated (TCM) scans.
• This method overcomes limitations of current CT dose metrics by integrating additional variables for more precise patient dose calculation.
• The GLM model shows particular promise for improving dose estimates in organs with high inter-patient variability.