GridAggregator

[MirenLurBarquin]

Hi all,

in TorchIO documentation there is an example of how to go back to the initial volume after using a GridSampler by using tio.inference.GridAggregator. In this example we are just working with one subject.

How could we go back to the initial volumes if we have a tio.data.SubjectsDataset containing multiple objects? For example after using a 'tio.Queue' and 'torch.utils.data()'. Is there any better way of doing it than with a dictionary of aggregators and creating subject_ids?

Thank you in advance. Best :)

[fepegar]

Hi, @MirenLurBarquin. The aggregator is for inference; the queue is for training. Can you please elaborate? What are you trying to do?

[MirenLurBarquin]

Hi @fepegar, thank you for your help.

I am doing axon segmentation. Sorry, I mixed train and inference on my first attempt. I think now I realized how should I do it: after training, if I want to do inference and visualize the results in more than one subject, I just do inference and use the aggregator for each of them one at a time, right? Following the example in the documentation.

[fepegar]

That sounds right. Please go through the tutorials and the documentation :)