Automatic characterization of thoracic aortic aneurysms from CT images

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Image Analysis

Background:

• Accurate thoracic aorta diameter measurement is crucial for diagnosing and monitoring aortic abnormalities like aneurysms.
• Existing methods may lack automation or robustness in diverse clinical scenarios.

Purpose of the Study:

• To develop and validate an automatic method for measuring thoracic aorta diameter.
• To assess the accuracy and rotational invariance of the proposed measurement technique.

Main Methods:

• Principal Component Analysis (PCA) applied to normal planes of the aorta to define its main axis.
• Validation using a database of 15 clinical cases compared against commercial software (gold standard).
• Rotational invariance tested using phantom images in various orientations.

Main Results:

• Good agreement observed between the automated method and the gold standard in clinical cases, with a bias of 0.2mm.
• Bland-Altman plots confirmed values within acceptable limits of agreement.
• ANOVA test showed no statistically significant difference in results for phantom images at different orientations (F=1.88, p=0.153), demonstrating robustness.

Conclusions:

• The developed automatic method provides accurate and reliable thoracic aorta diameter measurements.
• The technique exhibits robustness to rotational variations, making it suitable for clinical application.
• This approach can aid in the automatic characterization of thoracic aortic aneurysms.