NA-MIC Project Weeks

Back to Projects List

Extracting deep learning features from CT images of the thoracic region for lung cancer applications

Key Investigators

Renzo Phellan Aro (Lunenfeld-Tanenbaum Research Institute, Canada)
Martin Bellehumeur (Radical Imaging, Germany)

Project Description

This project focuses on using existing foundational deep learning models in 3D Slicer to process CT images of the thoracic region. The objective is to extract features that can be used for lung cancer applications, such as classification, screening, and risk prediction.

Objective

The main objective is to create a 3D Slicer application that receives either one or multiple CT images, processes them with a deep learning model, and generates features that can be used for various lung cancer applications.

Approach and Plan

Identify members of the 3D Slicer community interested in participating in the project.
Meet and discuss the scope of the project during the 3D Slicer week preparation meetings.
List the functional and GUI requirements for the application.
Identify useful deep learning foundational models for lung cancer applications.
Develop the 3D Slicer application to extract features from CT images of the thoracic region.

Progress

Worked on identifying members of the 3D Slicer community interested in participating in the project.
Used Claude AI to generate the graphical interface in 3D Slicer to process medical images using the Tangerine model.
Created a stand-alone command-line interface to access the model via OHIF.
Uploaded the code to a Github repo.

Next Steps

Look for more model models that can encode CT images to add to the interface.
Connect VS studio code via ssh to a server, for easier development with Claude AI.

Illustrations

Figure 1: Example screenshot showing the application interface in 3D Slicer.

Figure 2: Example screenshot showing the application interface in OHIF.