Automatic classification of hepatocellular carcinoma images based on nuclear and structural features

Tomoharu Kiyuna, Akira Saito, Atsushi Marugame, Yoshiko Yamashita, Maki Ogura, Eric Cosatto, Tokiya Abe, Akinori Hashiguchi, Michiie Sakamoto

Research output: Chapter in Book/Report/Conference proceedingConference contribution

6 Citations (Scopus)

Abstract

Diagnosis of hepatocellular carcinoma (HCC) on the basis of digital images is a challenging problem because, unlike gastrointestinal carcinoma, strong structural and morphological features are limited and sometimes absent from HCC images. In this study, we describe the classification of HCC images using statistical distributions of features obtained from image analysis of cell nuclei and hepatic trabeculae. Images of 130 hematoxylin-eosin (HE) stained histologic slides were captured at 20X by a slide scanner (Nanozoomer, Hamamatsu Photonics, Japan) and 1112 regions of interest (ROI) images were extracted for classification (551 negatives and 561 positives, including 113 well-differentiated positives). For a single nucleus, the following features were computed: area, perimeter, circularity, ellipticity, long and short axes of elliptic fit, contour complexity and gray level cooccurrence matrix (GLCM) texture features (angular second moment, contrast, homogeneity and entropy). In addition, distributions of nuclear density and hepatic trabecula thickness within an ROI were also extracted. To represent an ROI, statistical distributions (mean, standard deviation and percentiles) of these features were used. In total, 78 features were extracted for each ROI and a support vector machine (SVM) was trained to classify negative and positive ROIs. Experimental results using 5-fold cross validation show 90% sensitivity for an 87.8% specificity. The use of statistical distributions over a relatively large area makes the HCC classifier robust to occasional failures in the extraction of nuclear or hepatic trabecula features, thus providing stability to the system.

Original languageEnglish
Title of host publicationProgress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume8676
DOIs
Publication statusPublished - 2013
EventSPIE Medical Imaging Symposium 2013: Digital Pathology - Lake Buena Vista, FL, United States
Duration: 2013 Feb 102013 Feb 11

Other

OtherSPIE Medical Imaging Symposium 2013: Digital Pathology
CountryUnited States
CityLake Buena Vista, FL
Period13/2/1013/2/11

Fingerprint

Statistical Distributions
Hepatocellular Carcinoma
cancer
statistical distributions
Hematoxylin
Eosine Yellowish-(YS)
Photonics
Image analysis
Support vector machines
Liver
Classifiers
Entropy
chutes
Textures
Cells
Optics and Photonics
Cell Nucleus
nuclei
ellipticity
Japan

Keywords

  • Circle packing
  • Contour complexity texture
  • Digital pathology
  • GLCM
  • HCC
  • Support vector machine
  • Trabecula thickness

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics
  • Electronic, Optical and Magnetic Materials
  • Biomaterials
  • Radiology Nuclear Medicine and imaging

Cite this

Kiyuna, T., Saito, A., Marugame, A., Yamashita, Y., Ogura, M., Cosatto, E., ... Sakamoto, M. (2013). Automatic classification of hepatocellular carcinoma images based on nuclear and structural features. In Progress in Biomedical Optics and Imaging - Proceedings of SPIE (Vol. 8676). [86760Y] https://doi.org/10.1117/12.2006667

Automatic classification of hepatocellular carcinoma images based on nuclear and structural features. / Kiyuna, Tomoharu; Saito, Akira; Marugame, Atsushi; Yamashita, Yoshiko; Ogura, Maki; Cosatto, Eric; Abe, Tokiya; Hashiguchi, Akinori; Sakamoto, Michiie.

Progress in Biomedical Optics and Imaging - Proceedings of SPIE. Vol. 8676 2013. 86760Y.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Kiyuna, T, Saito, A, Marugame, A, Yamashita, Y, Ogura, M, Cosatto, E, Abe, T, Hashiguchi, A & Sakamoto, M 2013, Automatic classification of hepatocellular carcinoma images based on nuclear and structural features. in Progress in Biomedical Optics and Imaging - Proceedings of SPIE. vol. 8676, 86760Y, SPIE Medical Imaging Symposium 2013: Digital Pathology, Lake Buena Vista, FL, United States, 13/2/10. https://doi.org/10.1117/12.2006667
Kiyuna T, Saito A, Marugame A, Yamashita Y, Ogura M, Cosatto E et al. Automatic classification of hepatocellular carcinoma images based on nuclear and structural features. In Progress in Biomedical Optics and Imaging - Proceedings of SPIE. Vol. 8676. 2013. 86760Y https://doi.org/10.1117/12.2006667
Kiyuna, Tomoharu ; Saito, Akira ; Marugame, Atsushi ; Yamashita, Yoshiko ; Ogura, Maki ; Cosatto, Eric ; Abe, Tokiya ; Hashiguchi, Akinori ; Sakamoto, Michiie. / Automatic classification of hepatocellular carcinoma images based on nuclear and structural features. Progress in Biomedical Optics and Imaging - Proceedings of SPIE. Vol. 8676 2013.
@inproceedings{7d561fda5bc6477483e3a23be4bac7cc,
title = "Automatic classification of hepatocellular carcinoma images based on nuclear and structural features",
abstract = "Diagnosis of hepatocellular carcinoma (HCC) on the basis of digital images is a challenging problem because, unlike gastrointestinal carcinoma, strong structural and morphological features are limited and sometimes absent from HCC images. In this study, we describe the classification of HCC images using statistical distributions of features obtained from image analysis of cell nuclei and hepatic trabeculae. Images of 130 hematoxylin-eosin (HE) stained histologic slides were captured at 20X by a slide scanner (Nanozoomer, Hamamatsu Photonics, Japan) and 1112 regions of interest (ROI) images were extracted for classification (551 negatives and 561 positives, including 113 well-differentiated positives). For a single nucleus, the following features were computed: area, perimeter, circularity, ellipticity, long and short axes of elliptic fit, contour complexity and gray level cooccurrence matrix (GLCM) texture features (angular second moment, contrast, homogeneity and entropy). In addition, distributions of nuclear density and hepatic trabecula thickness within an ROI were also extracted. To represent an ROI, statistical distributions (mean, standard deviation and percentiles) of these features were used. In total, 78 features were extracted for each ROI and a support vector machine (SVM) was trained to classify negative and positive ROIs. Experimental results using 5-fold cross validation show 90{\%} sensitivity for an 87.8{\%} specificity. The use of statistical distributions over a relatively large area makes the HCC classifier robust to occasional failures in the extraction of nuclear or hepatic trabecula features, thus providing stability to the system.",
keywords = "Circle packing, Contour complexity texture, Digital pathology, GLCM, HCC, Support vector machine, Trabecula thickness",
author = "Tomoharu Kiyuna and Akira Saito and Atsushi Marugame and Yoshiko Yamashita and Maki Ogura and Eric Cosatto and Tokiya Abe and Akinori Hashiguchi and Michiie Sakamoto",
year = "2013",
doi = "10.1117/12.2006667",
language = "English",
isbn = "9780819494504",
volume = "8676",
booktitle = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",

}

TY - GEN

T1 - Automatic classification of hepatocellular carcinoma images based on nuclear and structural features

AU - Kiyuna, Tomoharu

AU - Saito, Akira

AU - Marugame, Atsushi

AU - Yamashita, Yoshiko

AU - Ogura, Maki

AU - Cosatto, Eric

AU - Abe, Tokiya

AU - Hashiguchi, Akinori

AU - Sakamoto, Michiie

PY - 2013

Y1 - 2013

N2 - Diagnosis of hepatocellular carcinoma (HCC) on the basis of digital images is a challenging problem because, unlike gastrointestinal carcinoma, strong structural and morphological features are limited and sometimes absent from HCC images. In this study, we describe the classification of HCC images using statistical distributions of features obtained from image analysis of cell nuclei and hepatic trabeculae. Images of 130 hematoxylin-eosin (HE) stained histologic slides were captured at 20X by a slide scanner (Nanozoomer, Hamamatsu Photonics, Japan) and 1112 regions of interest (ROI) images were extracted for classification (551 negatives and 561 positives, including 113 well-differentiated positives). For a single nucleus, the following features were computed: area, perimeter, circularity, ellipticity, long and short axes of elliptic fit, contour complexity and gray level cooccurrence matrix (GLCM) texture features (angular second moment, contrast, homogeneity and entropy). In addition, distributions of nuclear density and hepatic trabecula thickness within an ROI were also extracted. To represent an ROI, statistical distributions (mean, standard deviation and percentiles) of these features were used. In total, 78 features were extracted for each ROI and a support vector machine (SVM) was trained to classify negative and positive ROIs. Experimental results using 5-fold cross validation show 90% sensitivity for an 87.8% specificity. The use of statistical distributions over a relatively large area makes the HCC classifier robust to occasional failures in the extraction of nuclear or hepatic trabecula features, thus providing stability to the system.

AB - Diagnosis of hepatocellular carcinoma (HCC) on the basis of digital images is a challenging problem because, unlike gastrointestinal carcinoma, strong structural and morphological features are limited and sometimes absent from HCC images. In this study, we describe the classification of HCC images using statistical distributions of features obtained from image analysis of cell nuclei and hepatic trabeculae. Images of 130 hematoxylin-eosin (HE) stained histologic slides were captured at 20X by a slide scanner (Nanozoomer, Hamamatsu Photonics, Japan) and 1112 regions of interest (ROI) images were extracted for classification (551 negatives and 561 positives, including 113 well-differentiated positives). For a single nucleus, the following features were computed: area, perimeter, circularity, ellipticity, long and short axes of elliptic fit, contour complexity and gray level cooccurrence matrix (GLCM) texture features (angular second moment, contrast, homogeneity and entropy). In addition, distributions of nuclear density and hepatic trabecula thickness within an ROI were also extracted. To represent an ROI, statistical distributions (mean, standard deviation and percentiles) of these features were used. In total, 78 features were extracted for each ROI and a support vector machine (SVM) was trained to classify negative and positive ROIs. Experimental results using 5-fold cross validation show 90% sensitivity for an 87.8% specificity. The use of statistical distributions over a relatively large area makes the HCC classifier robust to occasional failures in the extraction of nuclear or hepatic trabecula features, thus providing stability to the system.

KW - Circle packing

KW - Contour complexity texture

KW - Digital pathology

KW - GLCM

KW - HCC

KW - Support vector machine

KW - Trabecula thickness

UR - http://www.scopus.com/inward/record.url?scp=84878575060&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84878575060&partnerID=8YFLogxK

U2 - 10.1117/12.2006667

DO - 10.1117/12.2006667

M3 - Conference contribution

SN - 9780819494504

VL - 8676

BT - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

ER -