Related Experiment Video
Updated: Sep 27, 2025
Ultrasound Imaging of the Thoracic and Abdominal Aorta in Mice to Determine Aneurysm Dimensions
Published on: March 8, 2019
Evaluating the Performance of a Convolutional Neural Network Algorithm for Measuring Thoracic Aortic Diameters in a
Caterina B Monti1, Marly van Assen1, Arthur E Stillman1
1Division of Cardiothoracic Imaging, Nuclear Medicine and Molecular Imaging, Department of Radiology and Imaging Sciences, Emory University Hospital, 1364 Clifton Rd NE, Atlanta, GA 30322 (C.B.M., M.v.A., A.E.S., S.J.L., C.N.D.C.); Department of Biomedical Sciences for Health, Università degli Studi di Milano, Milan, Italy (C.B.M., F. Secchi, F. Sardanelli); Digital Health Imaging Decision Support, Siemens Healthineers, Princeton, NJ (P.H.); Computed Tomography, Siemens Healthineers, Malvern, Pa (G.S.K.F.); and Unit of Radiology, Istituto di Ricovero e Cura a Carattere Scientifico Policlinico San Donato, San Donato Milanese, Italy (F. Secchi, F. Sardanelli).
A convolutional neural network (CNN) accurately measures thoracic aortic diameters in diverse CT scans. This AI tool shows good performance across various patient conditions and imaging equipment.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Cardiovascular Imaging
Background:
- Accurate thoracic aorta measurements are crucial for diagnosing and managing cardiovascular diseases.
- Manual measurements can be time-consuming and subject to inter-observer variability.
- Automated measurement tools using artificial intelligence show promise in improving efficiency and consistency.
Purpose of the Study:
- To evaluate the performance of a convolutional neural network (CNN) for automated thoracic aortic diameter measurements.
- To assess the accuracy and reproducibility of the CNN compared to manual measurements in a heterogeneous population.
- To identify factors influencing the performance of the automated measurement system.
Main Methods:
- Retrospective analysis of 233 chest CT scans from a heterogeneous population.
- Manual and CNN-based (Artificial Intelligence Rad Companion Chest CT prototype, Siemens Healthineers) measurements of aortic diameters at nine locations and maximum diameter.
- Bland-Altman analysis to assess agreement between manual and automatic measurements.
Main Results:
- No significant difference was observed in maximum aortic diameter between manual and automatic measurements (P = .48).
- Overall measurements showed a bias of -1.5 mm and a coefficient of repeatability of 8.0 mm.
- Factors such as contrast enhancement, aortic pathology (dissection, repair), and positioning affected reproducibility.
Conclusions:
- The CNN demonstrated good performance in measuring thoracic aortic diameters within a heterogeneous, multivendor CT dataset.
- Automated measurements offer a viable alternative to manual measurements, with potential for improved efficiency.
- Further refinement may be needed to address variability in specific patient subgroups and imaging conditions.
