predict the lung function of pulmonary fibrosis patients using pre diagnostic CT scan and longitudenal data of spirometer readings
- Extracting demographic information
- Handcrafting features using clinical information
- Generating features from a deep neural network
- Combining all the three types of features and building a regression model
- Testing and Evalaution
Segmentation pipline to extact lungs from CT scan and extracting Textural GLCM features :
Extracting latent features from neural network and visualzing the gradients - to underastand what the latent features represent :
Regression evaluation:
Model performance - demonstracted on two patients with different rate of decline:
