Related Experiment Video
Updated: Apr 13, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.7K
Automated scout-image-based estimation of contrast agent dosing: a deep learning approach
Robin Tibor Schirrmeister1, Laetitia Taleb2, Paul Friemel2
1Division of Medical Physics, Department of Diagnostic and Interventional Radiology, Medical Center - University of Freiburg, Faculty of Medicine - University of Freiburg, Freiburg, Germany.
BMC Medical Imaging
|April 11, 2026
Summary
A new deep learning algorithm automatically estimates patient weight from CT scout images, improving contrast agent dosing accuracy. This AI tool enhances patient safety and clinical efficiency in computed tomography (CT) workflows.
Area of Science:
- Radiology
- Artificial Intelligence
- Medical Imaging
Background:
- Accurate patient weight is crucial for optimal contrast agent dosing in computed tomography (CT).
- Manual weight measurements and self-reporting introduce workload and potential bias.
- Automating weight estimation from existing CT scout images is needed.
Purpose of the Study:
- To develop and validate a deep-learning algorithm for automatic patient weight estimation from CT scout images.
- To enable accurate contrast agent dosage calculation directly from imaging data.
- To improve efficiency and patient safety in CT procedures.
Main Methods:
- Retrospective analysis of 817 patients undergoing thorax/abdomen CT.
- Development of an EfficientNet convolutional neural network pipeline to estimate weight from scout images.
- Utilized in-context learning and dataset distillation for feature analysis; integrated into a browser-based UI.
Main Results:
- Self-reported weights were significantly lower than manual measurements (p < 10^-5).
- The pipeline achieved a mean absolute error (MAE) of 3.90 ± 0.20 kg in predicting patient weight.
- Predictive features included anatomical shape and overall image attenuation.
Conclusions:
- An open-source deep learning pipeline facilitates automatic and accurate contrast agent dosing in CT.
- The approach enhances patient safety and clinical efficiency by eliminating manual measurements.
- Further validation on larger cohorts and diverse clinical settings is recommended.
