Multiple classifier systems in texton-based approach for the classification of CT images of Lung

TY - GEN

T1 - Multiple classifier systems in texton-based approach for the classification of CT images of Lung

AU - Gangeh, Mehrdad J.

AU - Sørensen, Lauge

AU - Shaker, Saher B.

AU - Kamel, Mohamed S.

AU - de Bruijne, Marleen

PY - 2011

Y1 - 2011

N2 - In this paper, we propose using texton signatures based on raw pixelrepresentation along with a parallel multiple classifier system for the classification of emphysema in computed tomography images of the lung. The multiple classifier system is composed of support vector machines on the texton signatures as base classifiers and combines their decisions using product rule. The proposed approach is tested on 168 annotated regions of interest consisting of normal tissue, centrilobular emphysema, and paraseptal emphysema. Textonbased approach in texture classification mainly has two parameters, i.e., texton size and k value in k-means. Our results show that while aggregation of single decisions by SVMs over various k values using multiple classifier systems helps to improve the results compared to single SVMs, combining over differenttexton sizes is not beneficial. The performance of the proposed system, with anaccuracy of 95%, is similar to a recently proposed approach based on localbinary patterns, which performs almost the best among other approaches in theliterature.

AB - In this paper, we propose using texton signatures based on raw pixelrepresentation along with a parallel multiple classifier system for the classification of emphysema in computed tomography images of the lung. The multiple classifier system is composed of support vector machines on the texton signatures as base classifiers and combines their decisions using product rule. The proposed approach is tested on 168 annotated regions of interest consisting of normal tissue, centrilobular emphysema, and paraseptal emphysema. Textonbased approach in texture classification mainly has two parameters, i.e., texton size and k value in k-means. Our results show that while aggregation of single decisions by SVMs over various k values using multiple classifier systems helps to improve the results compared to single SVMs, combining over differenttexton sizes is not beneficial. The performance of the proposed system, with anaccuracy of 95%, is similar to a recently proposed approach based on localbinary patterns, which performs almost the best among other approaches in theliterature.

U2 - 10.1007/978-3-642-18421-5_15

DO - 10.1007/978-3-642-18421-5_15

M3 - Article in proceedings

SN - 978-3-642-18420-8

T3 - Lecture notes in computer science

SP - 153

EP - 163

BT - Medical Computer Vision. Recognition Techniques and Applications in Medical Imaging

A2 - Menze, Bjoern

A2 - Langs, Georg

A2 - Tu, Zhowen

A2 - Criminsi, Antonio

PB - Springer

T2 - Medical Computer Vision 2010: Recognition Techniques and Applications in Medical Imaging

Y2 - 20 September 2010 through 20 September 2010

ER -