Development of a new body weight estimation method using head CT scout images

Area of Science:

• Radiology and Medical Imaging
• Artificial Intelligence in Healthcare
• Emergency Medicine

Background:

• Accurate patient weight is essential for safe and effective imaging examinations, particularly in diagnosing acute ischemic stroke.
• Traditional weight measurement can be challenging and time-consuming in emergency settings.
• Head computed tomography (CT) scout images offer a potential source for rapid body weight estimation.

Purpose of the Study:

• To develop and evaluate a novel method for estimating body weight using head CT scout images.
• To facilitate contrast-enhanced CT examinations for acute ischemic stroke patients when direct weight measurement is difficult.

Main Methods:

• Investigated three weight estimation techniques: total pixel values from head CT scout images, the Xception deep learning model, and an average of these two methods.
• The Xception model was trained using 216 images with leave-one-out cross-validation.
• The primary focus was on the accuracy of the combined (average) weight estimation method.

Main Results:

• The novel combined method achieved moderate accuracy, with a 95% confidence interval of ±14.7 kg.
• The pixel value method had a wider interval (±20.6 kg), and the deep learning method had an interval of ±16.3 kg.
• The combined approach demonstrated superior performance compared to individual methods.

Conclusions:

• The developed weight estimation method, averaging two techniques, serves as a valuable support tool for medical staff in emergency situations.
• It enables rapid and feasible weight estimation for patients with acute conditions like stroke, where accurate measurements may not be readily obtainable.
• This technique can enhance the safety and efficiency of diagnostic imaging in critical care scenarios.