60 318

Cited 0 times in

Cited 2 times in

Evaluating brain volume segmentation accuracy and reliability of FreeSurfer and Neurophet AQUA at variations in MRI magnetic field strengths

Authors
 Lee, Hyunji  ;  Kim, Hye Weon  ;  Lee, Minho  ;  Kang, Jimin  ;  Kim, Donghyeon  ;  Lim, Hyun Kook  ;  Lee, Jun-Young  ;  Kim, Eosu  ;  Kim, Regina E. Y. 
Citation
 SCIENTIFIC REPORTS, Vol.14(1), 2024-10 
Article Number
 24513 
Journal Title
SCIENTIFIC REPORTS
ISSN
 2045-2322 
Issue Date
2024-10
Abstract
We aimed to compare the accuracy and reliability of two segmentation tools for magnetic resonance (MR) volumetry (FreeSurfer and Neurophet AQUA) at two magnetic field strengths (1.5T and 3T). We included 101 patients for the 1.5T-3T dataset and 112 for the 3T-3T dataset from three hospitals and five open-source datasets. The mean volume difference and average volume difference percentage with the change in magnetic field strength were compared between the methods. The hippocampus volume was larger with FreeSurfer than the Neurophet AQUA. In most brain regions, the Neurophet AQUA yielded a smaller average volume difference percentage (all < 10%) than FreeSurfer (all > 10%). The Neurophet AQUA exhibited more stable connectivity and regularity of the segmented components. Regarding volume, the Neurophet AQUA had effect sizes and ICCs comparable to those of FreeSurfer across the magnetic field strengths. With FreeSurfer, the original volume difference was small, whereas the average volume difference percentage was small with the Neurophet AQUA. Image segmentation took 1 h with FreeSurfer and 5 min with the Neurophet AQUA. When choosing an automatic segmentation method, the differences in image processing time and volume variability due to changes in the magnetic field strength of these methods should be considered.
DOI
10.1038/s41598-024-74622-y
Appears in Collections:
1. College of Medicine (의과대학) > Dept. of Psychiatry (정신과학교실) > 1. Journal Papers
Yonsei Authors
Kim, Eosu(김어수) ORCID logo https://orcid.org/0000-0001-9472-9465
URI
https://ir.ymlib.yonsei.ac.kr/handle/22282913/204518
사서에게 알리기
  feedback

qrcode

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

Browse

Links