NA-MIC Project Weeks
Back to Projects List
VolumeAXI
Key Investigators
- Jeanne CLARET (University of Michigan, USA)
- Gaëlle Leroux (University of Michigan, USA)
- Lucia Cevidanes (University of Michigan, USA)
- Claudia Mattos (University of Michigan, USA)
- Juan Prieto (University of North Carolina, USA)
Project Description
Volume Analysis, eXplainability and Interpretability, Volume-AXI, is an explainability approach for classification of bone and teeth structural defects in CBCT scans gray-level images. We propose to develop interpretable AI algorithms to visualize diagnostic features in dental and craniofacial conditions. This work is built on neural network models in Python, specifically using the MONAI framework,
The first clinical application of Volume-AXI is related to dentistry, aiming to identify the position of tooth impaction and damage to adjacente structures.
Objective
- Create AI algorithms capable of visualizing diagnostic features in dental and craniofacial conditions using CBCT (Cone Beam Computed Tomography) scan gray-level images.
- Integrate the developed AI algorithms with clinical workflows.
- Enhancing Explainability and Interpretability in Medical Imaging
Approach and Plan
-
Data Preparation and Pre-processing
-
Model Development and Training: Explore and select appropriate neural network architectures (e.g., CNNs, U-Nets) for image classification and feature visualization.
-
Explainability and Visualization Techniques: Implement methods to make AI decisions transparent and understandable such as Grad-CAM.
-
Validation and Testing
-
Documentation and Training: Create comprehensive documentation and user guides explaining the functionality and benefits of the AI tools.
Progress and Next Steps
- Done different preprocessing steps on the CBCT scans.
- Tried to train with EfficientNetBN.
Next step:
- Think about a new implementation of training.
- Try to reduce the image to regions of interest.
- Use of transformations in the training loop to increase the dataset.