Predicting standardized uptake value of brown adipose tissue from CT scans using convolutional neural networks

Ertunc Erdil (), Anton S. Becker, Moritz Schwyzer, Borja Martinez-Tellez, Jonatan R. Ruiz, Thomas Sartoretti, H. Alberto Vargas, A. Irene Burger, Alin Chirindel, Damian Wild, Nicola Zamboni, Bart Deplancke, Vincent Gardeux, Claudia Irene Maushart, Matthias Johannes Betz, Christian Wolfrum and Ender Konukoglu
Additional contact information
Ertunc Erdil: ETH Zurich
Anton S. Becker: ETH Zurich
Moritz Schwyzer: University Hospital Zurich
Borja Martinez-Tellez: University of Almería
Jonatan R. Ruiz: Sport and Health University Research Institute (iMUDS), University of Granada
Thomas Sartoretti: University Hospital Zurich
H. Alberto Vargas: Memorial Sloan Kettering Cancer Center
A. Irene Burger: University Zurich Hospital
Alin Chirindel: University Hospital of Basel
Damian Wild: University Hospital of Basel
Nicola Zamboni: ETH Zürich
Bart Deplancke: Ecole Polytechnique Fédérale de Lausanne (EPFL)
Vincent Gardeux: Ecole Polytechnique Fédérale de Lausanne (EPFL)
Claudia Irene Maushart: University Hospital Basel and University of Basel
Matthias Johannes Betz: University Hospital Basel and University of Basel
Christian Wolfrum: ETH Zurich
Ender Konukoglu: ETH Zurich

Nature Communications, 2024, vol. 15, issue 1, 1-14

Abstract: Abstract The standard method for identifying active Brown Adipose Tissue (BAT) is [18F]-Fluorodeoxyglucose ([18F]-FDG) PET/CT imaging, which is costly and exposes patients to radiation, making it impractical for population studies. These issues can be addressed with computational methods that predict [18F]-FDG uptake by BAT from CT; earlier population studies pave the way for developing such methods by showing some correlation between the Hounsfield Unit (HU) of BAT in CT and the corresponding [18F]-FDG uptake in PET. In this study, we propose training convolutional neural networks (CNNs) to predict [18F]-FDG uptake by BAT from unenhanced CT scans in the restricted regions that are likely to contain BAT. Using the Attention U-Net architecture, we perform experiments on datasets from four different cohorts, the largest study to date. We segment BAT regions using predicted [18F]-FDG uptake values, achieving 23% to 40% better accuracy than conventional CT thresholding. Additionally, BAT volumes computed from the segmentations distinguish the subjects with and without active BAT with an AUC of 0.8, compared to 0.6 for CT thresholding. These findings suggest CNNs can facilitate large-scale imaging studies more efficiently and cost-effectively using only CT.

Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.nature.com/articles/s41467-024-52622-w Abstract (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:nat:natcom:v:15:y:2024:i:1:d:10.1038_s41467-024-52622-w

Ordering information: This journal article can be ordered from
https://www.nature.com/ncomms/

DOI: 10.1038/s41467-024-52622-w

Access Statistics for this article

Nature Communications is currently edited by Nathalie Le Bot, Enda Bergin and Fiona Gillespie

More articles in Nature Communications from Nature
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().