Optimising brain age estimation through transfer learning: A suite of pre-trained foundation models for improved performance and generalisability in a clinical setting.

"we removed non-brain tissue from all 560 axial t 2 -weighted scans in the ixi dataset using hd-bet (isensee et al ) a publicly available deep learning-based skull-stripping tool accessible at https://github com/mic-dkfz/hd-bet .; scripts to enable readers to run and fine-tune our trained baseline models using their own mri scans are available at https://github com/midiconsortium/brainage 3 3 1 all baseline models representing the five commonest sequences and orientations in study dataset predicted chronological age with high accuracy in the internal clinical testing datasets (mae <= 4 0 years pearson's correlation r >= 93)."

"we removed non-brain tissue from all 560 axial t 2 -weighted scans in the ixi dataset using hd-bet (isensee et al ) a publicly available deep learning-based skull-stripping tool accessible at https://github com/mic-dkfz/hd-bet .; scripts to enable readers to run and fine-tune our trained baseline models using their own mri scans are available at https://github com/midiconsortium/brainage 3 3 1 all baseline models representing the five commonest sequences and orientations in study dataset predicted chronological age with high accuracy in the internal clinical testing datasets (mae <= 4 0 years pearson's correlation r >= 93).; scripts to enable readers to run our trained brain age models using their own scans are available at https://github com/midiconsortium/brainage ."

"CONFLICT OF INTEREST STATEMENT Co‐author Sebastien Ourselin is the co‐founder of Brainminer; however, he did not control or analyse the data. The other authors of this manuscript declare no relationships with any companies whose products or services may be related to the subject matter of the article."

"FUNDING INFORMATION This work was supported by the Royal College of Radiologists, King's College Hospital Research and Innovation, King's Health Partners Challenge Fund, NVIDIA (through the unrestricted use of a GPU obtained in a competition), the Wellcome/Engineering and Physical Sciences Research Council Center for Medical Engineering (WT 203148/Z/16/Z), and an MRC DPFS grant (MR/W021684/1)."