Premium
Harmonized benchmark labels of the hippocampus on magnetic resonance: The EADC‐ADNI project
Author(s) -
Bocchetta Martina,
Boccardi Marina,
Ganzola Rossana,
Apostolova Liana G.,
Preboske Gregory,
Wolf Dominik,
Ferrari Clarissa,
Pasqualetti Patrizio,
Robitaille Nicolas,
Duchesne Simon,
Jack Clifford R.,
Frisoni Giovanni B.,
Bartzokis George,
DeCarli Charles,
deToledoMorrell Leyla,
Fellgiebel Andreas,
Firbank Michael,
Gerritsen Lotte,
Henneman Wouter,
Killiany Ronald J.,
Malykhin Nikolai,
Pruessner Jens C.,
Soininen Hilkka,
Wang Lei,
Watson Craig,
Wolf Henrike
Publication year - 2015
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2013.12.019
Subject(s) - harp , magnetic resonance imaging , intraclass correlation , segmentation , benchmark (surveying) , neuroimaging , similarity (geometry) , artificial intelligence , computer science , confidence interval , pattern recognition (psychology) , medicine , psychology , neuroscience , radiology , cartography , developmental psychology , physics , psychometrics , quantum mechanics , image (mathematics) , geography
Background A globally harmonized protocol (HarP) for manual hippocampal segmentation based on magnetic resonance has been recently developed by a task force from European Alzheimer's Disease Consortium (EADC) and Alzheimer's Disease Neuroimaging Initiative (ADNI). Our aim was to produce benchmark labels based on the HarP for manual segmentation. Methods Five experts of manual hippocampal segmentation underwent specific training on the HarP and segmented 40 right and left hippocampi from 10 ADNI subjects on both 1.5 T and 3 T scans. An independent expert visually checked segmentations for compliance with the HarP. Descriptive measures of agreement between tracers were intraclass correlation coefficients (ICCs) of crude volumes and similarity coefficients of three‐dimensional volumes. Results Two hundred labels have been provided for the 20 magnetic resonance images. Intra‐ and interrater ICCs were >0.94, and mean similarity coefficients were 1.5 T, 0.73 (95% confidence interval [CI], 0.71–0.75); 3 T, 0.75 (95% CI, 0.74–0.76). Conclusion Certified benchmark labels have been produced based on the HarP to be used for tracers' training and qualification.