Automatic classification of white regions in liver biopsies by supervised machine learning

Scott Vanderbeck, Joseph Bockhorst, Richard Komorowski, David E. Kleiner, Samer Gawrieh

Research output: Contribution to journalArticle

15 Citations (Scopus)

Abstract

Automated assessment of histological features of non-alcoholic fatty liver disease (NAFLD) may reduce human variability and provide continuous rather than semiquantitative measurement of these features. As part of a larger effort, we perform automatic classification of steatosis, the cardinal feature of NAFLD, and other regions that manifest as white in images of hematoxylin and eosin-stained liver biopsy sections. These regions include macrosteatosis, central veins, portal veins, portal arteries, sinusoids and bile ducts. Digital images of hematoxylin and eosin-stained slides of 47 liver biopsies from patients with normal liver histology (n = 20) and NAFLD (n = 27) were obtained at 20× magnification. The images were analyzed using supervised machine learning classifiers created from annotations provided by two expert pathologists. The classification algorithm performs with 89% overall accuracy. It identified macrosteatosis, bile ducts, portal veins and sinusoids with high precision and recall (≥82%). Identification of central veins and portal arteries was less robust but still good. The accuracy of the classifier in identifying macrosteatosis is the best reported. The accurate automated identification of macrosteatosis achieved with this algorithm has useful clinical and research-related applications. The accurate detection of liver microscopic anatomical landmarks may facilitate important subsequent tasks, such as localization of other histological lesions according to liver microscopic anatomy.

Original languageEnglish (US)
Pages (from-to)785-792
Number of pages8
JournalHuman Pathology
Volume45
Issue number4
DOIs
StatePublished - 2014
Externally publishedYes

Fingerprint

Biopsy
Portal Vein
Liver
Hematoxylin
Eosine Yellowish-(YS)
Bile Ducts
Arteries
Veins
Anatomy
Histology
Supervised Machine Learning
Research
Non-alcoholic Fatty Liver Disease

Keywords

  • Digital image analysis
  • NAFLD
  • Sensitivity and specificity
  • Steatosis
  • Variability

ASJC Scopus subject areas

  • Pathology and Forensic Medicine

Cite this

Automatic classification of white regions in liver biopsies by supervised machine learning. / Vanderbeck, Scott; Bockhorst, Joseph; Komorowski, Richard; Kleiner, David E.; Gawrieh, Samer.

In: Human Pathology, Vol. 45, No. 4, 2014, p. 785-792.

Research output: Contribution to journalArticle

Vanderbeck, Scott ; Bockhorst, Joseph ; Komorowski, Richard ; Kleiner, David E. ; Gawrieh, Samer. / Automatic classification of white regions in liver biopsies by supervised machine learning. In: Human Pathology. 2014 ; Vol. 45, No. 4. pp. 785-792.
@article{2502e6a556d6492db4cfd39768fadafa,
title = "Automatic classification of white regions in liver biopsies by supervised machine learning",
abstract = "Automated assessment of histological features of non-alcoholic fatty liver disease (NAFLD) may reduce human variability and provide continuous rather than semiquantitative measurement of these features. As part of a larger effort, we perform automatic classification of steatosis, the cardinal feature of NAFLD, and other regions that manifest as white in images of hematoxylin and eosin-stained liver biopsy sections. These regions include macrosteatosis, central veins, portal veins, portal arteries, sinusoids and bile ducts. Digital images of hematoxylin and eosin-stained slides of 47 liver biopsies from patients with normal liver histology (n = 20) and NAFLD (n = 27) were obtained at 20× magnification. The images were analyzed using supervised machine learning classifiers created from annotations provided by two expert pathologists. The classification algorithm performs with 89{\%} overall accuracy. It identified macrosteatosis, bile ducts, portal veins and sinusoids with high precision and recall (≥82{\%}). Identification of central veins and portal arteries was less robust but still good. The accuracy of the classifier in identifying macrosteatosis is the best reported. The accurate automated identification of macrosteatosis achieved with this algorithm has useful clinical and research-related applications. The accurate detection of liver microscopic anatomical landmarks may facilitate important subsequent tasks, such as localization of other histological lesions according to liver microscopic anatomy.",
keywords = "Digital image analysis, NAFLD, Sensitivity and specificity, Steatosis, Variability",
author = "Scott Vanderbeck and Joseph Bockhorst and Richard Komorowski and Kleiner, {David E.} and Samer Gawrieh",
year = "2014",
doi = "10.1016/j.humpath.2013.11.011",
language = "English (US)",
volume = "45",
pages = "785--792",
journal = "Human Pathology",
issn = "0046-8177",
publisher = "W.B. Saunders Ltd",
number = "4",

}

TY - JOUR

T1 - Automatic classification of white regions in liver biopsies by supervised machine learning

AU - Vanderbeck, Scott

AU - Bockhorst, Joseph

AU - Komorowski, Richard

AU - Kleiner, David E.

AU - Gawrieh, Samer

PY - 2014

Y1 - 2014

N2 - Automated assessment of histological features of non-alcoholic fatty liver disease (NAFLD) may reduce human variability and provide continuous rather than semiquantitative measurement of these features. As part of a larger effort, we perform automatic classification of steatosis, the cardinal feature of NAFLD, and other regions that manifest as white in images of hematoxylin and eosin-stained liver biopsy sections. These regions include macrosteatosis, central veins, portal veins, portal arteries, sinusoids and bile ducts. Digital images of hematoxylin and eosin-stained slides of 47 liver biopsies from patients with normal liver histology (n = 20) and NAFLD (n = 27) were obtained at 20× magnification. The images were analyzed using supervised machine learning classifiers created from annotations provided by two expert pathologists. The classification algorithm performs with 89% overall accuracy. It identified macrosteatosis, bile ducts, portal veins and sinusoids with high precision and recall (≥82%). Identification of central veins and portal arteries was less robust but still good. The accuracy of the classifier in identifying macrosteatosis is the best reported. The accurate automated identification of macrosteatosis achieved with this algorithm has useful clinical and research-related applications. The accurate detection of liver microscopic anatomical landmarks may facilitate important subsequent tasks, such as localization of other histological lesions according to liver microscopic anatomy.

AB - Automated assessment of histological features of non-alcoholic fatty liver disease (NAFLD) may reduce human variability and provide continuous rather than semiquantitative measurement of these features. As part of a larger effort, we perform automatic classification of steatosis, the cardinal feature of NAFLD, and other regions that manifest as white in images of hematoxylin and eosin-stained liver biopsy sections. These regions include macrosteatosis, central veins, portal veins, portal arteries, sinusoids and bile ducts. Digital images of hematoxylin and eosin-stained slides of 47 liver biopsies from patients with normal liver histology (n = 20) and NAFLD (n = 27) were obtained at 20× magnification. The images were analyzed using supervised machine learning classifiers created from annotations provided by two expert pathologists. The classification algorithm performs with 89% overall accuracy. It identified macrosteatosis, bile ducts, portal veins and sinusoids with high precision and recall (≥82%). Identification of central veins and portal arteries was less robust but still good. The accuracy of the classifier in identifying macrosteatosis is the best reported. The accurate automated identification of macrosteatosis achieved with this algorithm has useful clinical and research-related applications. The accurate detection of liver microscopic anatomical landmarks may facilitate important subsequent tasks, such as localization of other histological lesions according to liver microscopic anatomy.

KW - Digital image analysis

KW - NAFLD

KW - Sensitivity and specificity

KW - Steatosis

KW - Variability

UR - http://www.scopus.com/inward/record.url?scp=84896542077&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84896542077&partnerID=8YFLogxK

U2 - 10.1016/j.humpath.2013.11.011

DO - 10.1016/j.humpath.2013.11.011

M3 - Article

C2 - 24565203

AN - SCOPUS:84896542077

VL - 45

SP - 785

EP - 792

JO - Human Pathology

JF - Human Pathology

SN - 0046-8177

IS - 4

ER -