Automatizing the Deep Learning pipeline for Out-of-Distribution Lung Segmentation

Fuente: Zenodo
Salvato in:
Dettagli Bibliografici
Autori principali: Adri Gómez, Monica, Abella
Natura: Recurso digital
Pubblicazione: Zenodo 2026
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866901531482128384
author Adri Gómez
Monica, Abella
author_facet Adri Gómez
Monica, Abella
contents <p>Data and models used in the paper "Automatizing the Deep Learning pipeline for Out-of-Distribution Lung Segmentation”, by A. Gómez, M. Buccardi, N. Miguel-Ramiro, ML. Soto-Montenegro, F. F. Stellari, M. Desco and M. Abella submitted to Scientific Reports, corresponding to the database SSA. </p> <p><strong>data.zip: </strong>Database consisting of 58 C57bl/6 rodent studies with varying levels of acute lung injury and their corresponding manual lungs segmentations. Input volumes are 512×512×576 voxels with an isotropic voxel size of 0.122 mm, whereas the voxel size for labels is doubled to 0.244 mm. The SSA folder must be dragged onto the "data" folder in AutoLung to run succesfully.</p> <p>The TCD and IST databases can be accessed through <a href="https://doi.org/10.6084/m9.figshare.c.4224377">https://doi.org/10.6084/m9.figshare.c.4224377</a> and the LUNA database can be accessed through <a href="https://luna16.grand-challenge.org/Data/">https://luna16.grand-challenge.org/Data/</a>. The QGX2 database is proprietary; to request access, please contact <a href="mailto:FB.Stellari@chiesi.com">FB.Stellari@chiesi.com</a>.</p> <p><strong>models.zip: </strong>Models trained on the joint SSA, IST and TCD dabasases, leaving 15% of the volumes for validation. The "localization" and "segmentation" folders, containing five pretrained models, must be dragged onto the "models" folder in AutoLung to run succesfully.</p> <p>The code necessary to reproduce these experiments, as well as a user manual for the segmentation interface, is available under the GPL-3.0 license at <a href="https://github.com/BiiGXray/AutoLung">https://github.com/BiiGXray/AutoLung</a>.</p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_19606321
institution Zenodo
language
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle Automatizing the Deep Learning pipeline for Out-of-Distribution Lung Segmentation
Adri Gómez
Monica, Abella
<p>Data and models used in the paper "Automatizing the Deep Learning pipeline for Out-of-Distribution Lung Segmentation”, by A. Gómez, M. Buccardi, N. Miguel-Ramiro, ML. Soto-Montenegro, F. F. Stellari, M. Desco and M. Abella submitted to Scientific Reports, corresponding to the database SSA. </p> <p><strong>data.zip: </strong>Database consisting of 58 C57bl/6 rodent studies with varying levels of acute lung injury and their corresponding manual lungs segmentations. Input volumes are 512×512×576 voxels with an isotropic voxel size of 0.122 mm, whereas the voxel size for labels is doubled to 0.244 mm. The SSA folder must be dragged onto the "data" folder in AutoLung to run succesfully.</p> <p>The TCD and IST databases can be accessed through <a href="https://doi.org/10.6084/m9.figshare.c.4224377">https://doi.org/10.6084/m9.figshare.c.4224377</a> and the LUNA database can be accessed through <a href="https://luna16.grand-challenge.org/Data/">https://luna16.grand-challenge.org/Data/</a>. The QGX2 database is proprietary; to request access, please contact <a href="mailto:FB.Stellari@chiesi.com">FB.Stellari@chiesi.com</a>.</p> <p><strong>models.zip: </strong>Models trained on the joint SSA, IST and TCD dabasases, leaving 15% of the volumes for validation. The "localization" and "segmentation" folders, containing five pretrained models, must be dragged onto the "models" folder in AutoLung to run succesfully.</p> <p>The code necessary to reproduce these experiments, as well as a user manual for the segmentation interface, is available under the GPL-3.0 license at <a href="https://github.com/BiiGXray/AutoLung">https://github.com/BiiGXray/AutoLung</a>.</p>
title Automatizing the Deep Learning pipeline for Out-of-Distribution Lung Segmentation
url https://doi.org/10.5281/zenodo.19606321