Measure DAE baseline with model with stronger DA

[JdbermeoUZH]

DoD:

• [x] Modify Nico's script to include logging and reporting of a specific set of labels of interest
• [ ] Modify my TTA object to measure/log specific labels of interest
• [x] Measured TTA DAE for both val and test with Nico's script for both brain and wmh
• [x] Measured TTA DAE for val with my script

Blocker:

• [x] The intensity normalization step before the regularizer has been fixed

[JdbermeoUZH]

Check what the normalizer is doing when it fails

• what is x_norm
• what does it predict?
• Why is dice nan, gt must have at least one mask of ones, unless it is discounting somehow the foreground