Deep Learning-Based Aortic Diameter Measurement in Traumatic Hemorrhage Using Shallow Attention Network: A Path...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Trauma Surgery

Background:

• Accurate aortic diameter (AoD) assessment is crucial for managing traumatic hemorrhage and guiding interventions like resuscitative endovascular balloon occlusion of the aorta (REBOA).
• Manual AoD measurements are time-consuming and prone to inter-observer variability, impacting clinical decision-making.
• Developing automated methods for AoD measurement can enhance efficiency and consistency in trauma care.

Purpose of the Study:

• To develop and validate a deep learning (DL) model for automated AoD measurement in trauma patients.
• To assess the model's accuracy and reliability compared to manual measurements.
• To evaluate the potential of DL in facilitating timely aortic interventions.

Main Methods:

• Retrospective analysis of abdominal CT scans from 300 adult trauma patients.
• Training a Shallow Attention Network on 444 manually annotated CT images for aortic segmentation and diameter measurement.
• Employing an ellipse-based calibration method to enhance measurement accuracy.

Main Results:

• The DL model achieved a high segmentation performance with a mean Dice coefficient of 0.865.
• Post-calibration, the mean discrepancy between predicted and ground truth diameters was minimal at 2.11 mm.
• The median diaphragmatic AoD was measured at 22.59 mm.

Conclusions:

• The developed deep learning model with ellipse-based calibration demonstrates robust performance for automated AoD measurement.
• This automated approach shows promise for improving the efficiency and accuracy of aortic diameter assessment in trauma patients.
• The model may aid in the timely planning of aortic interventions, such as REBOA, in critical care settings.