Computer aided diagnosis of diffuse lung disease in multi-detector CT - Selecting 3D texture features

I. Mariolis, Panagiotis Korfiatis, C. Kalogeropoulou, D. Daoussis, T. Petsas, L. Costaridou

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

4 Citations (Scopus)

Abstract

Computed Tomography (CT) is the modality of choice for the diagnosis of Diffuse Lung Disease (DLD) affecting lung parenchyma. The need for Computer Aided Diagnosis (CAD) schemes aimed at DLD patterns quantification in lung CT, originates from large inter- and intra observer variability characterizing DLD interpretation. The majority of the proposed CAD schemes aimed at DLD characterization exploits textural features combined with supervised classification algorithms. However, the exploitation of these features is suboptimal, since no feature reduction or evaluation is performed prior to the classification task. The aim of the current paper is to investigate 3D texture features sets (histogram signatures, co-occurrence and run length matrices' statistics) regarding their capability in DLD patterns' characterization (normal, ground glass, reticular and honeycombing). Earth Mover's Distance (EMD), k-Nearest Neighbor (k-NN) and Multinomial Logistic Regression (MLR) classifiers where used to access the performance of individual feature sets. In the analysis performed Histogram Signature (HS) feature set combined with EMD classifier, achieves the lowest overall accuracy (80.2 %). Co-occurrence based feature set presented the highest overall classification accuracy (99.3 %) when combined with k-NN classifier. However, both Run Length and Co-occurrence based feature sets, presented robustness against classifier choice and higher classification accuracy than HS feature set.

Original languageEnglish (US)
Title of host publicationXII Mediterranean Conference on Medical and Biological Engineering and Computing 2010, MEDICON 2010
Pages208-211
Number of pages4
DOIs
StatePublished - Oct 11 2010
Externally publishedYes
Event12th Mediterranean Conference on Medical and Biological Engineering and Computing, MEDICON 2010 - Chalkidiki, Greece
Duration: May 27 2010May 30 2010

Publication series

NameIFMBE Proceedings
Volume29
ISSN (Print)1680-0737

Conference

Conference12th Mediterranean Conference on Medical and Biological Engineering and Computing, MEDICON 2010
CountryGreece
CityChalkidiki
Period5/27/105/30/10

Fingerprint

Multidetector computed tomography
Computer aided diagnosis
Pulmonary diseases
Textures
Classifiers
Tomography
Earth (planet)
Logistics
Statistics
Glass

Keywords

  • 3D texture
  • Diffuse lung disease
  • Histogram signatures
  • MDCT
  • Supervised classification

ASJC Scopus subject areas

  • Biomedical Engineering
  • Bioengineering

Cite this

Mariolis, I., Korfiatis, P., Kalogeropoulou, C., Daoussis, D., Petsas, T., & Costaridou, L. (2010). Computer aided diagnosis of diffuse lung disease in multi-detector CT - Selecting 3D texture features. In XII Mediterranean Conference on Medical and Biological Engineering and Computing 2010, MEDICON 2010 (pp. 208-211). (IFMBE Proceedings; Vol. 29). https://doi.org/10.1007/978-3-642-13039-7_52

Computer aided diagnosis of diffuse lung disease in multi-detector CT - Selecting 3D texture features. / Mariolis, I.; Korfiatis, Panagiotis; Kalogeropoulou, C.; Daoussis, D.; Petsas, T.; Costaridou, L.

XII Mediterranean Conference on Medical and Biological Engineering and Computing 2010, MEDICON 2010. 2010. p. 208-211 (IFMBE Proceedings; Vol. 29).

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

Mariolis, I, Korfiatis, P, Kalogeropoulou, C, Daoussis, D, Petsas, T & Costaridou, L 2010, Computer aided diagnosis of diffuse lung disease in multi-detector CT - Selecting 3D texture features. in XII Mediterranean Conference on Medical and Biological Engineering and Computing 2010, MEDICON 2010. IFMBE Proceedings, vol. 29, pp. 208-211, 12th Mediterranean Conference on Medical and Biological Engineering and Computing, MEDICON 2010, Chalkidiki, Greece, 5/27/10. https://doi.org/10.1007/978-3-642-13039-7_52
Mariolis I, Korfiatis P, Kalogeropoulou C, Daoussis D, Petsas T, Costaridou L. Computer aided diagnosis of diffuse lung disease in multi-detector CT - Selecting 3D texture features. In XII Mediterranean Conference on Medical and Biological Engineering and Computing 2010, MEDICON 2010. 2010. p. 208-211. (IFMBE Proceedings). https://doi.org/10.1007/978-3-642-13039-7_52
Mariolis, I. ; Korfiatis, Panagiotis ; Kalogeropoulou, C. ; Daoussis, D. ; Petsas, T. ; Costaridou, L. / Computer aided diagnosis of diffuse lung disease in multi-detector CT - Selecting 3D texture features. XII Mediterranean Conference on Medical and Biological Engineering and Computing 2010, MEDICON 2010. 2010. pp. 208-211 (IFMBE Proceedings).
@inproceedings{50db35e4543e44f5b3e27c4760ba2ff6,
title = "Computer aided diagnosis of diffuse lung disease in multi-detector CT - Selecting 3D texture features",
abstract = "Computed Tomography (CT) is the modality of choice for the diagnosis of Diffuse Lung Disease (DLD) affecting lung parenchyma. The need for Computer Aided Diagnosis (CAD) schemes aimed at DLD patterns quantification in lung CT, originates from large inter- and intra observer variability characterizing DLD interpretation. The majority of the proposed CAD schemes aimed at DLD characterization exploits textural features combined with supervised classification algorithms. However, the exploitation of these features is suboptimal, since no feature reduction or evaluation is performed prior to the classification task. The aim of the current paper is to investigate 3D texture features sets (histogram signatures, co-occurrence and run length matrices' statistics) regarding their capability in DLD patterns' characterization (normal, ground glass, reticular and honeycombing). Earth Mover's Distance (EMD), k-Nearest Neighbor (k-NN) and Multinomial Logistic Regression (MLR) classifiers where used to access the performance of individual feature sets. In the analysis performed Histogram Signature (HS) feature set combined with EMD classifier, achieves the lowest overall accuracy (80.2 {\%}). Co-occurrence based feature set presented the highest overall classification accuracy (99.3 {\%}) when combined with k-NN classifier. However, both Run Length and Co-occurrence based feature sets, presented robustness against classifier choice and higher classification accuracy than HS feature set.",
keywords = "3D texture, Diffuse lung disease, Histogram signatures, MDCT, Supervised classification",
author = "I. Mariolis and Panagiotis Korfiatis and C. Kalogeropoulou and D. Daoussis and T. Petsas and L. Costaridou",
year = "2010",
month = "10",
day = "11",
doi = "10.1007/978-3-642-13039-7_52",
language = "English (US)",
isbn = "9783642130380",
series = "IFMBE Proceedings",
pages = "208--211",
booktitle = "XII Mediterranean Conference on Medical and Biological Engineering and Computing 2010, MEDICON 2010",

}

TY - GEN

T1 - Computer aided diagnosis of diffuse lung disease in multi-detector CT - Selecting 3D texture features

AU - Mariolis, I.

AU - Korfiatis, Panagiotis

AU - Kalogeropoulou, C.

AU - Daoussis, D.

AU - Petsas, T.

AU - Costaridou, L.

PY - 2010/10/11

Y1 - 2010/10/11

N2 - Computed Tomography (CT) is the modality of choice for the diagnosis of Diffuse Lung Disease (DLD) affecting lung parenchyma. The need for Computer Aided Diagnosis (CAD) schemes aimed at DLD patterns quantification in lung CT, originates from large inter- and intra observer variability characterizing DLD interpretation. The majority of the proposed CAD schemes aimed at DLD characterization exploits textural features combined with supervised classification algorithms. However, the exploitation of these features is suboptimal, since no feature reduction or evaluation is performed prior to the classification task. The aim of the current paper is to investigate 3D texture features sets (histogram signatures, co-occurrence and run length matrices' statistics) regarding their capability in DLD patterns' characterization (normal, ground glass, reticular and honeycombing). Earth Mover's Distance (EMD), k-Nearest Neighbor (k-NN) and Multinomial Logistic Regression (MLR) classifiers where used to access the performance of individual feature sets. In the analysis performed Histogram Signature (HS) feature set combined with EMD classifier, achieves the lowest overall accuracy (80.2 %). Co-occurrence based feature set presented the highest overall classification accuracy (99.3 %) when combined with k-NN classifier. However, both Run Length and Co-occurrence based feature sets, presented robustness against classifier choice and higher classification accuracy than HS feature set.

AB - Computed Tomography (CT) is the modality of choice for the diagnosis of Diffuse Lung Disease (DLD) affecting lung parenchyma. The need for Computer Aided Diagnosis (CAD) schemes aimed at DLD patterns quantification in lung CT, originates from large inter- and intra observer variability characterizing DLD interpretation. The majority of the proposed CAD schemes aimed at DLD characterization exploits textural features combined with supervised classification algorithms. However, the exploitation of these features is suboptimal, since no feature reduction or evaluation is performed prior to the classification task. The aim of the current paper is to investigate 3D texture features sets (histogram signatures, co-occurrence and run length matrices' statistics) regarding their capability in DLD patterns' characterization (normal, ground glass, reticular and honeycombing). Earth Mover's Distance (EMD), k-Nearest Neighbor (k-NN) and Multinomial Logistic Regression (MLR) classifiers where used to access the performance of individual feature sets. In the analysis performed Histogram Signature (HS) feature set combined with EMD classifier, achieves the lowest overall accuracy (80.2 %). Co-occurrence based feature set presented the highest overall classification accuracy (99.3 %) when combined with k-NN classifier. However, both Run Length and Co-occurrence based feature sets, presented robustness against classifier choice and higher classification accuracy than HS feature set.

KW - 3D texture

KW - Diffuse lung disease

KW - Histogram signatures

KW - MDCT

KW - Supervised classification

UR - http://www.scopus.com/inward/record.url?scp=77957607660&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77957607660&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-13039-7_52

DO - 10.1007/978-3-642-13039-7_52

M3 - Conference contribution

AN - SCOPUS:77957607660

SN - 9783642130380

T3 - IFMBE Proceedings

SP - 208

EP - 211

BT - XII Mediterranean Conference on Medical and Biological Engineering and Computing 2010, MEDICON 2010

ER -