Finally out!! Our latest segmentation tool, called LOD-Brain has been published in Medical Image Analysis!
It can segment any T1w (1.5T/3T) without atlas alignment nor preprocessing in less than a minute. More here: https://doi.org/10.1016/j.media.2024.103090
Around 27000 T1w volumes from ~160 sites in the world were used in the study…
Now, some results on 78 datasets (clinical, research, and synthetic MRIs):
Do you want to see something cool? Here results on five patients with only one brain hemisphere:
A working demo is also available here (click on demo). Just drop and drop a volume and see the result: https://rocknroll87q.github.io/LOD-Brain/
Code, model, and more results are available on the project website: https://rocknroll87q.github.io/LOD-Brain/
A big thanks to my co-authors: Mattia Savardi, Alberto Signoroni, Sergio Benini, and Lars Muckli.
And an acknowledgement to the Human Brain Project:
This was supported from the EU Horizon 2020 Framework Programme for Research and Innovation under the Specific Grant Agreement No. 945539 (Human Brain Project SGA3).