Fast fractal stack: fractal analysis of computed tomography scans of the lung

Authors:
Alceu Ferraz Costa;Joe Tekli;Agma Juci Machado Traina
Affiliations:
University of São Paulo, São Carlos, Brazil;University of São Paulo, São Carlos, Brazil;University of São Paulo, São Carlos, Brazil
Venue:
MMAR '11 Proceedings of the 2011 international ACM workshop on Medical multimedia analysis and retrieval
Year:
2011

Citing 6
Cited 0

Invariant Image Recognition by Zernike Moments

IEEE Transactions on Pattern Analysis and Machine Intelligence
Fast training of support vector machines using sequential minimal optimization

Advances in kernel methods
Correlation-based Feature Selection for Discrete and Numeric Class Machine Learning

ICML '00 Proceedings of the Seventeenth International Conference on Machine Learning
Fractal Analysis of Image Textures for Indexing and Retrieval by Content

CBMS '05 Proceedings of the 18th IEEE Symposium on Computer-Based Medical Systems
Clustering high-dimensional data: A survey on subspace clustering, pattern-based clustering, and correlation clustering

ACM Transactions on Knowledge Discovery from Data (TKDD)
Fusing visual and clinical information for lung tissue classification in high-resolution computed tomography

Artificial Intelligence in Medicine

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper proposes a new feature extraction method: the Fast Fractal Stack, or FFS. The extraction algorithm consists in decomposing the input grayscale image into a stack of binary images from which the fractal dimension values are computed, resulting in a compact and highly descriptive set of features. We evaluated FFS for the task of classification of interstitial lung diseases in computed tomography (CT) scans, applied on a database of 248 CT images from 67 patients. The proposed approach performs well, improving the classification accuracy when compared to other feature extraction algorithms. Additionally, the FFS extraction algorithm is efficient, with a computational cost linear with respect to input image size.