-
As mentioned in the title, this repository is evolving towards a collection of utility files for dataset conversion or training scripts or models, etc. Hence, I was thinking of renaming it to "utiliti…
-
### Description
When working on #4042, I edited the model description:
https://github.com/spinalcordtoolbox/spinalcordtoolbox/blob/c59867e436cb0b4bb5c3c23d9074ebf250e1b7a7/spinalcordtoolbox/deep…
-
As mentioned in #11, both me and @naga-karthik observe heavy overfitting in training for this project. I believe this might be because there is a very large variability in lesions across subjects whic…
-
Given the high performance of the nnUnet in general, it would be a good idea to have a benchmark with this architecture.
Todo (update if necessary):
- [x] Write a script to convert the current B…
-
Here is the followed strategy for the final model training for the zurich-mouse dataset for white and grey matter segmentation.
- [x] Extract annotated 2D slices from the dataset
- [x] Train a 2D…
-
To save time, I tried cropping only with a 5vox ball (instead of 5vox ball + 32vox disk as in the [processing](https://github.com/ivadomed/model_seg_ms_mp2rage/blob/33a3a717e9b9677622b57d09980d121bf5e…
-
The JSON files associated with the Zurich dataset are not BIDS compliant. They seem to have been generated with a Matlab software. Example for **sub-zh01_ses-01_acq-ax_T2w**:
```json
{
"history": …
-
The idea would be to train several models, with randomized train/validation split, and then aggregate the various inferences.
Possibly relevant: https://colab.research.google.com/github/Project-MO…
-
# Context / Description
When training several deep learning models in production (split 80/20/0) with different seeds, the idea is to average all predictions to generate a more reliable soft predic…
-
This issue aims to open a brief discussion about the type of registration method to be used for co-registering the axial and sagittal T2w scans as a prerequisite before training a multichannel model. …