Modify mapping function to mask out segments that are surplus in the label volume

neuronets / nobrainer

A framework for developing neural network models for 3D image processing.

Other

159 stars 45 forks source link

Modify mapping function to mask out segments that are surplus in the label volume #327

Closed hvgazula closed 7 months ago

hvgazula commented 7 months ago

Turns out that not all labels that are in the volume are present in the provided label mapping volumes. This means we have to identify the surplus labels non-eagerly and this could be more challenging if the surplus labels (across all volumes) are not known ahead of time.

hvgazula commented 7 months ago

https://github.com/neuronets/nobrainer/blob/976691d685824fd4bba836498abea4184cffd798/nobrainer/volume.py#L24 takes care of this.

hvgazula commented 7 months ago

snippet to calculate surplus labels in all volumes https://github.com/neuronets/nobrainer_training_scripts/commit/eaa6cf954fa78191035a08fba7373d0e6357e7a1