Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Clinical Application of Automatic Segmentation of Medial Temporal Lobe Subregions in Prodromal and Dementia-Level Alzheimer's Disease

Gertje, Eske Christiane LU orcid ; Pluta, John ; Das, Sandhitsu ; Mancuso, Lauren ; Kliot, Dasha ; Yushkevich, Paul and Wolk, David (2016) In Journal of Alzheimer's disease : JAD 54(3). p.1027-1037
Abstract

BACKGROUND: Volumetry of medial temporal lobe (MTL) structures to diagnose Alzheimer's disease (AD) in its earliest symptomatic stage could be of great importance for interventions or disease modifying pharmacotherapy.

OBJECTIVE: This study aimed to demonstrate the first application of an automatic segmentation method of MTL subregions in a clinical population. Automatic segmentation of magnetic resonance images (MRIs) in a research population has previously been shown to detect evidence of neurodegeneration in MTL subregions and to help discriminate AD and mild cognitive impairment (MCI) from a healthy comparison group.

METHODS: Clinical patients were selected and T2-weighted MRI scan quality was checked. An automatic... (More)

BACKGROUND: Volumetry of medial temporal lobe (MTL) structures to diagnose Alzheimer's disease (AD) in its earliest symptomatic stage could be of great importance for interventions or disease modifying pharmacotherapy.

OBJECTIVE: This study aimed to demonstrate the first application of an automatic segmentation method of MTL subregions in a clinical population. Automatic segmentation of magnetic resonance images (MRIs) in a research population has previously been shown to detect evidence of neurodegeneration in MTL subregions and to help discriminate AD and mild cognitive impairment (MCI) from a healthy comparison group.

METHODS: Clinical patients were selected and T2-weighted MRI scan quality was checked. An automatic segmentation method of hippocampal subfields (ASHS) was applied to scans of 67 AD patients, 38 amnestic MCI patients, and 57 healthy controls. Hippocampal subfields, entorhinal cortex (ERC), and perirhinal cortex were automatically labeled and subregion volumes were compared between groups.

RESULTS: One fourth of all scans were excluded due to bad scan quality. There were significant volume reductions in all subregions, except BA36, in aMCIs (p < 0.001), most prominently in Cornu Ammonis 1 (CA1) and ERC, and in all subregions in AD. However, sensitivity of CA1 and ERC hardly differed from sensitivity of WH in aMCI and AD.

CONCLUSION: Applying automatic segmentation of MTL subregions in a clinical setting as a potential biomarker for prodromal AD is feasible, but issues of image quality due to motion remain to be addressed. CA1 and ERC provided strongest group discrimination in differentiating aMCIs from controls, but discriminatory power of different subfields was low overall.

(Less)
Please use this url to cite or link to this publication:
author
; ; ; ; ; and
publishing date
type
Contribution to journal
publication status
published
subject
keywords
Aged, Aged, 80 and over, Alzheimer Disease/diagnostic imaging, Cognitive Dysfunction/diagnostic imaging, Cohort Studies, Dementia/diagnostic imaging, Female, Humans, Image Processing, Computer-Assisted/methods, Magnetic Resonance Imaging/methods, Male, Middle Aged, Prodromal Symptoms, Temporal Lobe/diagnostic imaging
in
Journal of Alzheimer's disease : JAD
volume
54
issue
3
pages
1027 - 1037
publisher
SAGE Publications
external identifiers
  • pmid:27567809
  • scopus:84990026693
ISSN
1387-2877
DOI
10.3233/JAD-160014
language
English
LU publication?
no
id
767e1028-b015-4914-ab54-ec5b9adae980
date added to LUP
2025-03-05 12:25:57
date last changed
2025-04-04 15:24:56
@article{767e1028-b015-4914-ab54-ec5b9adae980,
  abstract     = {{<p>BACKGROUND: Volumetry of medial temporal lobe (MTL) structures to diagnose Alzheimer's disease (AD) in its earliest symptomatic stage could be of great importance for interventions or disease modifying pharmacotherapy.</p><p>OBJECTIVE: This study aimed to demonstrate the first application of an automatic segmentation method of MTL subregions in a clinical population. Automatic segmentation of magnetic resonance images (MRIs) in a research population has previously been shown to detect evidence of neurodegeneration in MTL subregions and to help discriminate AD and mild cognitive impairment (MCI) from a healthy comparison group.</p><p>METHODS: Clinical patients were selected and T2-weighted MRI scan quality was checked. An automatic segmentation method of hippocampal subfields (ASHS) was applied to scans of 67 AD patients, 38 amnestic MCI patients, and 57 healthy controls. Hippocampal subfields, entorhinal cortex (ERC), and perirhinal cortex were automatically labeled and subregion volumes were compared between groups.</p><p>RESULTS: One fourth of all scans were excluded due to bad scan quality. There were significant volume reductions in all subregions, except BA36, in aMCIs (p &lt; 0.001), most prominently in Cornu Ammonis 1 (CA1) and ERC, and in all subregions in AD. However, sensitivity of CA1 and ERC hardly differed from sensitivity of WH in aMCI and AD.</p><p>CONCLUSION: Applying automatic segmentation of MTL subregions in a clinical setting as a potential biomarker for prodromal AD is feasible, but issues of image quality due to motion remain to be addressed. CA1 and ERC provided strongest group discrimination in differentiating aMCIs from controls, but discriminatory power of different subfields was low overall.</p>}},
  author       = {{Gertje, Eske Christiane and Pluta, John and Das, Sandhitsu and Mancuso, Lauren and Kliot, Dasha and Yushkevich, Paul and Wolk, David}},
  issn         = {{1387-2877}},
  keywords     = {{Aged; Aged, 80 and over; Alzheimer Disease/diagnostic imaging; Cognitive Dysfunction/diagnostic imaging; Cohort Studies; Dementia/diagnostic imaging; Female; Humans; Image Processing, Computer-Assisted/methods; Magnetic Resonance Imaging/methods; Male; Middle Aged; Prodromal Symptoms; Temporal Lobe/diagnostic imaging}},
  language     = {{eng}},
  month        = {{10}},
  number       = {{3}},
  pages        = {{1027--1037}},
  publisher    = {{SAGE Publications}},
  series       = {{Journal of Alzheimer's disease : JAD}},
  title        = {{Clinical Application of Automatic Segmentation of Medial Temporal Lobe Subregions in Prodromal and Dementia-Level Alzheimer's Disease}},
  url          = {{http://dx.doi.org/10.3233/JAD-160014}},
  doi          = {{10.3233/JAD-160014}},
  volume       = {{54}},
  year         = {{2016}},
}