AI Classifier for False Lumen Thrombosis After Aortic Surgery
Author Information
Author(s): Osswald Anja, Tsagakis Konstantinos, Thielmann Matthias, Lumsden Alan B., Ruhparwar Arjang, Karmonik Christof, Chen Dechang
Primary Institution: University Duisburg-Essen
Hypothesis
Can an unsupervised AI algorithm accurately identify and quantify false lumen thrombosis in CTA images after Frozen Elephant Trunk operations?
Conclusion
The AI algorithm can automatically classify the presence and absence of false lumen thrombosis in patient-specific CTA images.
Supporting Evidence
- The algorithm achieved high concordance between reconstructed and original images.
- Average thrombus scores varied significantly between patients with and without thrombus.
- The study utilized a large dataset of 3938 images for training the AI model.
- Automated classification could reduce the time and variability associated with manual analysis.
- Integration of the AI tool could enhance post-operative monitoring efficiency.
Takeaway
This study shows that a computer program can help doctors quickly tell if there's a problem with blood clots in patients' aorta after surgery.
Methodology
The study used CTA images from eight patients and trained a variational autoencoder (VAE) with a 2D convolutional neural network to classify false lumen thrombosis.
Limitations
The study primarily demonstrates feasibility; further research is needed to establish the algorithm's efficacy and limitations.
Participant Demographics
Patients underwent Frozen Elephant Trunk operations for aortic dissection, with varying degrees of false lumen thrombosis.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website