Clarification Request: Training Datasets Used for liver_segment, liver_vessels Models

[anuragp2018]

First of all, thank you for the excellent work on TotalSegmentator. We are currently using the following liver-related models in our research pipeline:

liver_segment_1

liver_segment_2

liver_segment_3

liver_segment_4

liver_segment_5

liver_segment_6

liver_segment_7

liver_segment_8

liver_vessels

liver_tumor

We understand that the main TotalSegmentator (104-structure) model was trained on the ~1200 CT dataset described in the Radiology AI publication.

However, we would like clarification regarding the liver-specific subtasks listed above:

1. Were these models trained on the same original TotalSegmentator CT dataset?

2. Or were separate task-specific datasets used?

3.If separate datasets were used:
Are they public datasets?
Could you please share the dataset names and links (if available)?

4.Are these models trained purely using nnU-Net with default configuration, or were any task-specific modifications applied?

We are preparing technical documentation and would like to accurately cite the correct training datasets and methodology.

Thank you for your time and clarification.