Tissue classification using cluster features for lesion detection in digital cervigrams

Sharon Xiaolei Huang, Wei Wang, Zhiyun Xue, Sameer Antani, L. Rodney Long, Jose Jeronimo

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

19 Citations (Scopus)

Abstract

In this paper, we propose a new method for automated detection and segmentation of different tissue types in digitized uterine cervix images using mean-shift clustering and support vector machines (SVM) classification on cluster features. We specifically target the segmentation of precancerous lesions in a NCI/NLM archive of 60,000 cervigrams. Due to large variations in image appearance in the archive, color and texture features of a tissue type in one image often overlap with that of a different tissue type in another image. This makes reliable tissue segmentation in a large number of images a very challenging problem. In this paper, we propose the use of powerful machine learning techniques such as Support Vector Machines (SVM) to learn, from a database with ground truth annotations, critical visual signs that correlate with important tissue types and to use the learned classifier for tissue segmentation in unseen images. In our experiments, SVM performs better than un-supervised methods such as Gaussian Mixture clustering, but it does not scale very well to large training sets and does not always guarantee improved performance given more training data. To address this problem, we combine SVM and clustering so that the features we extracted for classification are features of clusters returned by the mean-shift clustering algorithm. Compared to classification using individual pixel features, classification by cluster features greatly reduces the dimensionality of the problem, thus it is more efficient while producing results with comparable accuracy.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2008
Subtitle of host publicationImage Processing
DOIs
StatePublished - May 19 2008
EventMedical Imaging 2008: Image Processing - San Diego, CA, United States
Duration: Feb 17 2008Feb 19 2008

Publication series

NameProgress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume6914
ISSN (Print)1605-7422

Other

OtherMedical Imaging 2008: Image Processing
CountryUnited States
CitySan Diego, CA
Period2/17/082/19/08

Fingerprint

Tissue
Support vector machines
Clustering algorithms
Learning systems
Classifiers
Textures
Pixels
Color
Experiments

All Science Journal Classification (ASJC) codes

  • Engineering(all)

Cite this

Huang, S. X., Wang, W., Xue, Z., Antani, S., Long, L. R., & Jeronimo, J. (2008). Tissue classification using cluster features for lesion detection in digital cervigrams. In Medical Imaging 2008: Image Processing [69141Z] (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 6914). https://doi.org/10.1117/12.771088
Huang, Sharon Xiaolei ; Wang, Wei ; Xue, Zhiyun ; Antani, Sameer ; Long, L. Rodney ; Jeronimo, Jose. / Tissue classification using cluster features for lesion detection in digital cervigrams. Medical Imaging 2008: Image Processing. 2008. (Progress in Biomedical Optics and Imaging - Proceedings of SPIE).
@inproceedings{3859645635024d4eaafc6263290e378a,
title = "Tissue classification using cluster features for lesion detection in digital cervigrams",
abstract = "In this paper, we propose a new method for automated detection and segmentation of different tissue types in digitized uterine cervix images using mean-shift clustering and support vector machines (SVM) classification on cluster features. We specifically target the segmentation of precancerous lesions in a NCI/NLM archive of 60,000 cervigrams. Due to large variations in image appearance in the archive, color and texture features of a tissue type in one image often overlap with that of a different tissue type in another image. This makes reliable tissue segmentation in a large number of images a very challenging problem. In this paper, we propose the use of powerful machine learning techniques such as Support Vector Machines (SVM) to learn, from a database with ground truth annotations, critical visual signs that correlate with important tissue types and to use the learned classifier for tissue segmentation in unseen images. In our experiments, SVM performs better than un-supervised methods such as Gaussian Mixture clustering, but it does not scale very well to large training sets and does not always guarantee improved performance given more training data. To address this problem, we combine SVM and clustering so that the features we extracted for classification are features of clusters returned by the mean-shift clustering algorithm. Compared to classification using individual pixel features, classification by cluster features greatly reduces the dimensionality of the problem, thus it is more efficient while producing results with comparable accuracy.",
author = "Huang, {Sharon Xiaolei} and Wei Wang and Zhiyun Xue and Sameer Antani and Long, {L. Rodney} and Jose Jeronimo",
year = "2008",
month = "5",
day = "19",
doi = "10.1117/12.771088",
language = "English (US)",
isbn = "9780819470980",
series = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",
booktitle = "Medical Imaging 2008",

}

Huang, SX, Wang, W, Xue, Z, Antani, S, Long, LR & Jeronimo, J 2008, Tissue classification using cluster features for lesion detection in digital cervigrams. in Medical Imaging 2008: Image Processing., 69141Z, Progress in Biomedical Optics and Imaging - Proceedings of SPIE, vol. 6914, Medical Imaging 2008: Image Processing, San Diego, CA, United States, 2/17/08. https://doi.org/10.1117/12.771088

Tissue classification using cluster features for lesion detection in digital cervigrams. / Huang, Sharon Xiaolei; Wang, Wei; Xue, Zhiyun; Antani, Sameer; Long, L. Rodney; Jeronimo, Jose.

Medical Imaging 2008: Image Processing. 2008. 69141Z (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 6914).

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

TY - GEN

T1 - Tissue classification using cluster features for lesion detection in digital cervigrams

AU - Huang, Sharon Xiaolei

AU - Wang, Wei

AU - Xue, Zhiyun

AU - Antani, Sameer

AU - Long, L. Rodney

AU - Jeronimo, Jose

PY - 2008/5/19

Y1 - 2008/5/19

N2 - In this paper, we propose a new method for automated detection and segmentation of different tissue types in digitized uterine cervix images using mean-shift clustering and support vector machines (SVM) classification on cluster features. We specifically target the segmentation of precancerous lesions in a NCI/NLM archive of 60,000 cervigrams. Due to large variations in image appearance in the archive, color and texture features of a tissue type in one image often overlap with that of a different tissue type in another image. This makes reliable tissue segmentation in a large number of images a very challenging problem. In this paper, we propose the use of powerful machine learning techniques such as Support Vector Machines (SVM) to learn, from a database with ground truth annotations, critical visual signs that correlate with important tissue types and to use the learned classifier for tissue segmentation in unseen images. In our experiments, SVM performs better than un-supervised methods such as Gaussian Mixture clustering, but it does not scale very well to large training sets and does not always guarantee improved performance given more training data. To address this problem, we combine SVM and clustering so that the features we extracted for classification are features of clusters returned by the mean-shift clustering algorithm. Compared to classification using individual pixel features, classification by cluster features greatly reduces the dimensionality of the problem, thus it is more efficient while producing results with comparable accuracy.

AB - In this paper, we propose a new method for automated detection and segmentation of different tissue types in digitized uterine cervix images using mean-shift clustering and support vector machines (SVM) classification on cluster features. We specifically target the segmentation of precancerous lesions in a NCI/NLM archive of 60,000 cervigrams. Due to large variations in image appearance in the archive, color and texture features of a tissue type in one image often overlap with that of a different tissue type in another image. This makes reliable tissue segmentation in a large number of images a very challenging problem. In this paper, we propose the use of powerful machine learning techniques such as Support Vector Machines (SVM) to learn, from a database with ground truth annotations, critical visual signs that correlate with important tissue types and to use the learned classifier for tissue segmentation in unseen images. In our experiments, SVM performs better than un-supervised methods such as Gaussian Mixture clustering, but it does not scale very well to large training sets and does not always guarantee improved performance given more training data. To address this problem, we combine SVM and clustering so that the features we extracted for classification are features of clusters returned by the mean-shift clustering algorithm. Compared to classification using individual pixel features, classification by cluster features greatly reduces the dimensionality of the problem, thus it is more efficient while producing results with comparable accuracy.

UR - http://www.scopus.com/inward/record.url?scp=43449132708&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=43449132708&partnerID=8YFLogxK

U2 - 10.1117/12.771088

DO - 10.1117/12.771088

M3 - Conference contribution

SN - 9780819470980

T3 - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

BT - Medical Imaging 2008

ER -

Huang SX, Wang W, Xue Z, Antani S, Long LR, Jeronimo J. Tissue classification using cluster features for lesion detection in digital cervigrams. In Medical Imaging 2008: Image Processing. 2008. 69141Z. (Progress in Biomedical Optics and Imaging - Proceedings of SPIE). https://doi.org/10.1117/12.771088