Automated Characterization of Abdominal MRI Exams Using Deep Learning
2024
Automated Characterization of Abdominal MRI Exams Using Deep Learning
publication
Evidence: high
Author Information
Author(s): Kim Joonghyun, Chae Allison, Duda Jeffrey, Borthakur Ari, Rader Daniel, Gee James C., Kahn Charles E., Witschey Walter R., Sagreiya Hersh
Primary Institution: BioBank Penn Medicine
Hypothesis
Can convolutional neural networks effectively classify abdominal MRI sequences, orientations, and contrasts?
Conclusion
The study found that specialized CNN models can accurately classify abdominal MRI data with high performance.
Supporting Evidence
- The CNN models achieved accuracies of 96.9% for sequence classification, 97.4% for orientation, and 97.3% for contrast.
Takeaway
This study shows that computers can learn to recognize different types of MRI images really well, helping doctors understand them better.
Methodology
Three distinct CNN models were trained to classify MRI images into sequences, orientations, and contrasts.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website