Please enable JavaScript.

Coggle requires JavaScript to display documents.

Experimental Setup, Results and Discussion, Our Models, Folder structure…

- - - - Image residues
        
        wrt to diffrent existing models
        
        wrt to the baseline time stamp going as input
        
        jacobbian
    - - Tables comparison wrt evaluation metircs
    - - umap/tsne/ spectral of how original and shifted embeddings differ
  - - - segmentation maps of different parts
        
        Relative volume diff.--only csf
      - Apart from the view which is synthesized are the other views realistic is appearance?? coronal--sagital
- - - - ResNet/VGG
- - - - Penalizing the intensity of the images (increasing time point increasing intensity)
    - - The discriminator loss which penalizes wrong ordering
        
        This can be added to the R
    - - The discriminator loss takes the real image from a different subject containing to the same age
- - - - config_file_ADNI
        
        start slice = 50, end slice = 61, total slice = 181
        
        data key --------- MCI_M_train
      - config_file
      - config_file_ADNI_classifier
        
        data key --------- MCI_M_train
        
        start slice = 84, end slice = 96, total slice = 181
      - config_file_brats2020
  - - - sample ADNI MCI_M and makes a dictionary for each slice with all necessary info ----- original image, generated image, age, slice no, nii path ...
    - - after sampling the ADNI images and saving each slice as a dictionary this code will update the dictionary with some extra info----- other timestamp data
- - - - each slice of the test set of MCI_M is stored as a dictionary with the samples from the diffusion model along with all the original information
    - - each slice of the test set of MCI_M is stored as a dictionary with the samples from the diffusion model along with all the original information