Automated Characterization of Abdominal MRI Exams Using Deep Learning
Author Information
Author(s): Kim Joonghyun, Chae Allison, Duda Jeffrey, Borthakur Ari, Rader Daniel, Gee James C., Kahn Charles E. Jr, Witschey Walter R., Sagreiya Hersh
Primary Institution: University of Pennsylvania
Hypothesis
Can convolutional neural networks effectively classify abdominal MRI sequences, orientations, and contrast types?
Conclusion
The study demonstrates that convolutional neural networks can accurately classify abdominal MRI sequences, orientations, and contrast types with high accuracy.
Supporting Evidence
- The CNN models achieved accuracies of 96.9%, 97.4%, and 97.3% for sequence, orientation, and contrast classification, respectively.
- The study utilized a dataset of 29,361 images after removing corrupted files.
- Automated classification can enhance the efficiency and standardization of MRI interpretation across institutions.
Takeaway
This study shows that computers can learn to tell different types of MRI pictures apart, which helps doctors understand them better.
Methodology
The study used convolutional neural networks to classify abdominal MRI images into sequences, orientations, and contrast types.
Potential Biases
Potential misclassifications due to similarities in image features and the complexity of abdominal anatomy.
Limitations
The study's dataset was limited to a single institution, which may affect the generalizability of the findings.
Participant Demographics
Participants were from the Penn Medicine BioBank, but specific demographic details are not provided.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website