Related Experiment Video
Updated: Sep 22, 2025
07:30
3D Printing Model of a Patient's Specific Lumbar Vertebra
Published on: April 14, 2023
2.1K
Diagnosis of Lumbar Spondylolisthesis Using a Pruned CNN Model
Deepika Saravagi1, Shweta Agrawal2, Manisha Saravagi3
1Department of Computer Application, SAGE University, Indore 452012, India.
Computational and Mathematical Methods in Medicine
|May 20, 2022
Summary
This study optimized convolutional neural network (CNN) models for diagnosing lumbar spondylolisthesis on small devices. Unit pruning significantly reduced model complexity and improved diagnostic accuracy compared to weight pruning.
Area of Science:
- Medical imaging analysis
- Artificial intelligence in healthcare
Background:
- Convolutional neural network (CNN) models show significant progress in medical applications.
- High parameter counts in CNNs limit their use on resource-constrained devices.
Purpose of the Study:
- To reduce the complexity of CNN models for lumbar spondylolisthesis diagnosis.
- To enable the deployment of CNN models on small devices for medical diagnosis.
Main Methods:
- Applied weight and unit pruning strategies to reduce CNN model complexity.
- Evaluated the performance of pruned models for lumbar spondylolisthesis diagnosis.
Main Results:
- Unit pruning outperformed weight pruning, achieving 94.12% accuracy.
- Reduced network load by 90% through unit pruning.
- Significantly fewer parameters contributed to the outcome with unit pruning (10%) compared to weight pruning (30%).
Conclusions:
- The proposed pruned CNN model offers higher accuracy for lumbar spondylolisthesis diagnosis than previous models.
- Unit pruning is an effective strategy for optimizing CNNs for medical applications on small devices.
