Interpretable classification of Alzheimer’s disease pathologies with a convolutional neural network pipeline

Ziqi Tang, Kangway V. Chuang, Charles DeCarli, Lee-Way Jin, Laurel A Beckett, Michael J. Keiser, Brittany Dugger

Research output: Contribution to journalArticle

Abstract

Neuropathologists assess vast brain areas to identify diverse and subtly-differentiated morphologies. Standard semi-quantitative scoring approaches, however, are coarse-grained and lack precise neuroanatomic localization. We report a proof-of-concept deep learning pipeline that identifies specific neuropathologies—amyloid plaques and cerebral amyloid angiopathy—in immunohistochemically-stained archival slides. Using automated segmentation of stained objects and a cloud-based interface, we annotate > 70,000 plaque candidates from 43 whole slide images (WSIs) to train and evaluate convolutional neural networks. Networks achieve strong plaque classification on a 10-WSI hold-out set (0.993 and 0.743 areas under the receiver operating characteristic and precision recall curve, respectively). Prediction confidence maps visualize morphology distributions at high resolution. Resulting network-derived amyloid beta (Aβ)-burden scores correlate well with established semi-quantitative scores on a 30-WSI blinded hold-out. Finally, saliency mapping demonstrates that networks learn patterns agreeing with accepted pathologic features. This scalable means to augment a neuropathologist’s ability suggests a route to neuropathologic deep phenotyping.

Original languageEnglish (US)
Article number2173
JournalNature communications
Volume10
Issue number1
DOIs
StatePublished - Dec 1 2019

Fingerprint

pathology
Pathology
chutes
Amyloid
Alzheimer Disease
Pipelines
Neural networks
Aptitude
Amyloid Plaques
ROC Curve
Brain
Learning
scoring
learning
brain
confidence
receivers
routes
Neuropathology
high resolution

ASJC Scopus subject areas

  • Chemistry(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Physics and Astronomy(all)

Cite this

Interpretable classification of Alzheimer’s disease pathologies with a convolutional neural network pipeline. / Tang, Ziqi; Chuang, Kangway V.; DeCarli, Charles; Jin, Lee-Way; Beckett, Laurel A; Keiser, Michael J.; Dugger, Brittany.

In: Nature communications, Vol. 10, No. 1, 2173, 01.12.2019.

Research output: Contribution to journalArticle

@article{e736d08fe0df4a3c83d36c10b176db82,
title = "Interpretable classification of Alzheimer’s disease pathologies with a convolutional neural network pipeline",
abstract = "Neuropathologists assess vast brain areas to identify diverse and subtly-differentiated morphologies. Standard semi-quantitative scoring approaches, however, are coarse-grained and lack precise neuroanatomic localization. We report a proof-of-concept deep learning pipeline that identifies specific neuropathologies—amyloid plaques and cerebral amyloid angiopathy—in immunohistochemically-stained archival slides. Using automated segmentation of stained objects and a cloud-based interface, we annotate > 70,000 plaque candidates from 43 whole slide images (WSIs) to train and evaluate convolutional neural networks. Networks achieve strong plaque classification on a 10-WSI hold-out set (0.993 and 0.743 areas under the receiver operating characteristic and precision recall curve, respectively). Prediction confidence maps visualize morphology distributions at high resolution. Resulting network-derived amyloid beta (Aβ)-burden scores correlate well with established semi-quantitative scores on a 30-WSI blinded hold-out. Finally, saliency mapping demonstrates that networks learn patterns agreeing with accepted pathologic features. This scalable means to augment a neuropathologist’s ability suggests a route to neuropathologic deep phenotyping.",
author = "Ziqi Tang and Chuang, {Kangway V.} and Charles DeCarli and Lee-Way Jin and Beckett, {Laurel A} and Keiser, {Michael J.} and Brittany Dugger",
year = "2019",
month = "12",
day = "1",
doi = "10.1038/s41467-019-10212-1",
language = "English (US)",
volume = "10",
journal = "Nature Communications",
issn = "2041-1723",
publisher = "Nature Publishing Group",
number = "1",

}

TY - JOUR

T1 - Interpretable classification of Alzheimer’s disease pathologies with a convolutional neural network pipeline

AU - Tang, Ziqi

AU - Chuang, Kangway V.

AU - DeCarli, Charles

AU - Jin, Lee-Way

AU - Beckett, Laurel A

AU - Keiser, Michael J.

AU - Dugger, Brittany

PY - 2019/12/1

Y1 - 2019/12/1

N2 - Neuropathologists assess vast brain areas to identify diverse and subtly-differentiated morphologies. Standard semi-quantitative scoring approaches, however, are coarse-grained and lack precise neuroanatomic localization. We report a proof-of-concept deep learning pipeline that identifies specific neuropathologies—amyloid plaques and cerebral amyloid angiopathy—in immunohistochemically-stained archival slides. Using automated segmentation of stained objects and a cloud-based interface, we annotate > 70,000 plaque candidates from 43 whole slide images (WSIs) to train and evaluate convolutional neural networks. Networks achieve strong plaque classification on a 10-WSI hold-out set (0.993 and 0.743 areas under the receiver operating characteristic and precision recall curve, respectively). Prediction confidence maps visualize morphology distributions at high resolution. Resulting network-derived amyloid beta (Aβ)-burden scores correlate well with established semi-quantitative scores on a 30-WSI blinded hold-out. Finally, saliency mapping demonstrates that networks learn patterns agreeing with accepted pathologic features. This scalable means to augment a neuropathologist’s ability suggests a route to neuropathologic deep phenotyping.

AB - Neuropathologists assess vast brain areas to identify diverse and subtly-differentiated morphologies. Standard semi-quantitative scoring approaches, however, are coarse-grained and lack precise neuroanatomic localization. We report a proof-of-concept deep learning pipeline that identifies specific neuropathologies—amyloid plaques and cerebral amyloid angiopathy—in immunohistochemically-stained archival slides. Using automated segmentation of stained objects and a cloud-based interface, we annotate > 70,000 plaque candidates from 43 whole slide images (WSIs) to train and evaluate convolutional neural networks. Networks achieve strong plaque classification on a 10-WSI hold-out set (0.993 and 0.743 areas under the receiver operating characteristic and precision recall curve, respectively). Prediction confidence maps visualize morphology distributions at high resolution. Resulting network-derived amyloid beta (Aβ)-burden scores correlate well with established semi-quantitative scores on a 30-WSI blinded hold-out. Finally, saliency mapping demonstrates that networks learn patterns agreeing with accepted pathologic features. This scalable means to augment a neuropathologist’s ability suggests a route to neuropathologic deep phenotyping.

UR - http://www.scopus.com/inward/record.url?scp=85065794854&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85065794854&partnerID=8YFLogxK

U2 - 10.1038/s41467-019-10212-1

DO - 10.1038/s41467-019-10212-1

M3 - Article

VL - 10

JO - Nature Communications

JF - Nature Communications

SN - 2041-1723

IS - 1

M1 - 2173

ER -