Local histograms for classifying H&E stained tissues

M. L. Massar, R. Bhagavatula, M. Fickus, Jelena Kovacevic

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

Abstract

We introduce a rigorous mathematical theory for the analysis of local histograms, and consider the appropriateness of their use in the automated classification of textures commonly encountered in images of H&E stained tissues. We first discuss some of the many image features that pathologists indicate they use when classifying tissues, focusing on simple, locally-defined features that essentially involve pixel counting: the number of cells in a region of given size, the size of the nuclei within these cells, and the distribution of color within both. We then introduce a probabilistic, occlusion-based model for textures that exhibit these features, in particular demonstrating how certain tissue-similar textures can be built up from simpler ones. After considering the basic notions and properties of local histogram transforms, we then formally demonstrate that such transforms are natural tools for analyzing the textures produced by our model. In particular, we discuss how local histogram transforms can be used to produce numerical features that, when fed into mainstream classification schemes, mimic the baser aspects of a pathologist's thought process.

Original languageEnglish (US)
Title of host publication26th Southern Biomedical Engineering Conference SBEC 2010
Pages348-352
Number of pages5
Volume32 IFMBE
DOIs
StatePublished - Nov 8 2010
Event26th Southern Biomedical Engineering Conference, SBEC 2010 - College Park, MD, United States
Duration: Apr 30 2010May 2 2010

Other

Other26th Southern Biomedical Engineering Conference, SBEC 2010
CountryUnited States
CityCollege Park, MD
Period4/30/105/2/10

Fingerprint

Textures
Tissue
Pixels
Color

Keywords

  • histology
  • local histogram
  • occlusion

ASJC Scopus subject areas

  • Bioengineering
  • Biomedical Engineering

Cite this

Massar, M. L., Bhagavatula, R., Fickus, M., & Kovacevic, J. (2010). Local histograms for classifying H&E stained tissues. In 26th Southern Biomedical Engineering Conference SBEC 2010 (Vol. 32 IFMBE, pp. 348-352) https://doi.org/10.1007/978-3-642-14998-6_89

Local histograms for classifying H&E stained tissues. / Massar, M. L.; Bhagavatula, R.; Fickus, M.; Kovacevic, Jelena.

26th Southern Biomedical Engineering Conference SBEC 2010. Vol. 32 IFMBE 2010. p. 348-352.

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

Massar, ML, Bhagavatula, R, Fickus, M & Kovacevic, J 2010, Local histograms for classifying H&E stained tissues. in 26th Southern Biomedical Engineering Conference SBEC 2010. vol. 32 IFMBE, pp. 348-352, 26th Southern Biomedical Engineering Conference, SBEC 2010, College Park, MD, United States, 4/30/10. https://doi.org/10.1007/978-3-642-14998-6_89
Massar ML, Bhagavatula R, Fickus M, Kovacevic J. Local histograms for classifying H&E stained tissues. In 26th Southern Biomedical Engineering Conference SBEC 2010. Vol. 32 IFMBE. 2010. p. 348-352 https://doi.org/10.1007/978-3-642-14998-6_89
Massar, M. L. ; Bhagavatula, R. ; Fickus, M. ; Kovacevic, Jelena. / Local histograms for classifying H&E stained tissues. 26th Southern Biomedical Engineering Conference SBEC 2010. Vol. 32 IFMBE 2010. pp. 348-352
@inproceedings{fc716f4156a6477cbda0330af9c693c5,
title = "Local histograms for classifying H&E stained tissues",
abstract = "We introduce a rigorous mathematical theory for the analysis of local histograms, and consider the appropriateness of their use in the automated classification of textures commonly encountered in images of H&E stained tissues. We first discuss some of the many image features that pathologists indicate they use when classifying tissues, focusing on simple, locally-defined features that essentially involve pixel counting: the number of cells in a region of given size, the size of the nuclei within these cells, and the distribution of color within both. We then introduce a probabilistic, occlusion-based model for textures that exhibit these features, in particular demonstrating how certain tissue-similar textures can be built up from simpler ones. After considering the basic notions and properties of local histogram transforms, we then formally demonstrate that such transforms are natural tools for analyzing the textures produced by our model. In particular, we discuss how local histogram transforms can be used to produce numerical features that, when fed into mainstream classification schemes, mimic the baser aspects of a pathologist's thought process.",
keywords = "histology, local histogram, occlusion",
author = "Massar, {M. L.} and R. Bhagavatula and M. Fickus and Jelena Kovacevic",
year = "2010",
month = "11",
day = "8",
doi = "10.1007/978-3-642-14998-6_89",
language = "English (US)",
isbn = "9783642149979",
volume = "32 IFMBE",
pages = "348--352",
booktitle = "26th Southern Biomedical Engineering Conference SBEC 2010",

}

TY - GEN

T1 - Local histograms for classifying H&E stained tissues

AU - Massar, M. L.

AU - Bhagavatula, R.

AU - Fickus, M.

AU - Kovacevic, Jelena

PY - 2010/11/8

Y1 - 2010/11/8

N2 - We introduce a rigorous mathematical theory for the analysis of local histograms, and consider the appropriateness of their use in the automated classification of textures commonly encountered in images of H&E stained tissues. We first discuss some of the many image features that pathologists indicate they use when classifying tissues, focusing on simple, locally-defined features that essentially involve pixel counting: the number of cells in a region of given size, the size of the nuclei within these cells, and the distribution of color within both. We then introduce a probabilistic, occlusion-based model for textures that exhibit these features, in particular demonstrating how certain tissue-similar textures can be built up from simpler ones. After considering the basic notions and properties of local histogram transforms, we then formally demonstrate that such transforms are natural tools for analyzing the textures produced by our model. In particular, we discuss how local histogram transforms can be used to produce numerical features that, when fed into mainstream classification schemes, mimic the baser aspects of a pathologist's thought process.

AB - We introduce a rigorous mathematical theory for the analysis of local histograms, and consider the appropriateness of their use in the automated classification of textures commonly encountered in images of H&E stained tissues. We first discuss some of the many image features that pathologists indicate they use when classifying tissues, focusing on simple, locally-defined features that essentially involve pixel counting: the number of cells in a region of given size, the size of the nuclei within these cells, and the distribution of color within both. We then introduce a probabilistic, occlusion-based model for textures that exhibit these features, in particular demonstrating how certain tissue-similar textures can be built up from simpler ones. After considering the basic notions and properties of local histogram transforms, we then formally demonstrate that such transforms are natural tools for analyzing the textures produced by our model. In particular, we discuss how local histogram transforms can be used to produce numerical features that, when fed into mainstream classification schemes, mimic the baser aspects of a pathologist's thought process.

KW - histology

KW - local histogram

KW - occlusion

UR - http://www.scopus.com/inward/record.url?scp=78049396138&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78049396138&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-14998-6_89

DO - 10.1007/978-3-642-14998-6_89

M3 - Conference contribution

SN - 9783642149979

VL - 32 IFMBE

SP - 348

EP - 352

BT - 26th Southern Biomedical Engineering Conference SBEC 2010

ER -