Training labels for hippocampal segmentation based on the EADC-ADNI harmonized hippocampal protocol

Marina Boccardi, Martina Bocchetta, Félix C. Morency, D. Louis Collins, Masami Nishikawa, Rossana Ganzola, Michel J. Grothe, Dominik Wolf, Alberto Redolfi, Michela Pievani, Luigi Antelmi, Andreas Fellgiebel, Hiroshi Matsuda, Stefan Teipel, Simon Duchesne, Clifford R. Jack, Giovanni B. Frisoni

Research output: Contribution to journalArticlepeer-review

70 Scopus citations

Abstract

Background: The European Alzheimer's Disease Consortium and Alzheimer's Disease Neuroimaging Initiative (ADNI) Harmonized Protocol (HarP) is a Delphi definition of manual hippocampal segmentation from magnetic resonance imaging (MRI) that can be used as the standard of truth to train new tracers, and to validate automated segmentation algorithms. Training requires large and representative data sets of segmented hippocampi. This work aims to produce a set of HarP labels for the proper training and certification of tracers and algorithms. Methods: Sixty-eight 1.5 T and 67 3 T volumetric structural ADNI scans from different subjects, balanced by age, medial temporal atrophy, and scanner manufacturer, were segmented by five qualified HarP tracers whose absolute interrater intraclass correlation coefficients were 0.953 and 0.975 (left and right). Labels were validated as HarP compliant through centralized quality check and correction. Results: Hippocampal volumes (mm3) were as follows: controls: left 5 3060 (standard deviation [SD], 502), right 5 3120 (SD, 897); mild cognitive impairment (MCI): left 5 2596 (SD, 447), right 5 2686 (SD, 473); and Alzheimer's disease (AD): left 5 2301 (SD, 492), right 5 2445 (SD, 525). Volumes significantly correlated with atrophy severity at Scheltens' scale (Spearman's r5 ,20.468, P5 ,.0005). Background: The European Alzheimer's Disease Consortium and Alzheimer's Disease Neuroimaging Initiative (ADNI) Harmonized Protocol (HarP) is a Delphi definition of manual hippocampal segmentation from magnetic resonance imaging (MRI) that can be used as the standard of truth to train new tracers, and to validate automated segmentation algorithms. Training requires large and representative data sets of segmented hippocampi. This work aims to produce a set of HarP labels for the proper training and certification of tracers and algorithms. Methods: Sixty-eight 1.5 T and 67 3 T volumetric structural ADNI scans from different subjects, balanced by age, medial temporal atrophy, and scanner manufacturer, were segmented by five qualified HarP tracers whose absolute interrater intraclass correlation coefficients were 0.953 and 0.975 (left and right). Labels were validated as HarP compliant through centralized quality check and correction. Results: Hippocampal volumes (mm3) were as follows: controls: left 5 3060 (standard deviation [SD], 502), right 5 3120 (SD, 897); mild cognitive impairment (MCI): left 5 2596 (SD, 447), right 5 2686 (SD, 473); and Alzheimer's disease (AD): left 5 2301 (SD, 492), right 5 2445 (SD, 525). Volumes significantly correlated with atrophy severity at Scheltens' scale (Spearman's r5 ,20.468, P5 ,.0005).

Original languageEnglish (US)
Pages (from-to)175-183
Number of pages9
JournalAlzheimer's and Dementia
Volume11
Issue number2
DOIs
StatePublished - 2015

Keywords

  • Algorithm training
  • Automated segmentation algorithms
  • Benchmark images
  • Harmonized protocol
  • Hippocampal segmentation
  • Hippocampus
  • MRI

ASJC Scopus subject areas

  • Clinical Neurology
  • Geriatrics and Gerontology
  • Psychiatry and Mental health
  • Cellular and Molecular Neuroscience
  • Health Policy
  • Developmental Neuroscience
  • Epidemiology

Fingerprint

Dive into the research topics of 'Training labels for hippocampal segmentation based on the EADC-ADNI harmonized hippocampal protocol'. Together they form a unique fingerprint.

Cite this