Texture analysis of hepatocellular carcinoma and liver cysts in CT images

Authors:
Daniel Smutek;Akinobu Shimizu;Hidefumi Kobatake;Shigeru Nawano;Ludvik Tesar
Affiliations:
Tokyo University of Agriculture and Technology, Koganei, Tokyo, Japan and 1st Medical Faculty, Charles University, Prague, Czech Republic;Tokyo University of Agriculture and Technology, Koganei, Tokyo, Japan;Tokyo University of Agriculture and Technology, Koganei, Tokyo, Japan;National Cancer Center Hospital East, Kashiwa, Chiba, Japan;Inst. of Information Theory and Automation, Czech Academy of Sciences, Prague, Czech Republic
Venue:
SPPRA'06 Proceedings of the 24th IASTED international conference on Signal processing, pattern recognition, and applications
Year:
2006

Citing 4
Cited 1

Neural networks for pattern recognition

Neural networks for pattern recognition
On Combining Classifiers

IEEE Transactions on Pattern Analysis and Machine Intelligence
Information fusion in biomedical image analysis: combination of data vs. combination of interpretations

IPMI'05 Proceedings of the 19th international conference on Information Processing in Medical Imaging
A computer-aided diagnostic system to characterize CT focal liver lesions: design and optimization of a neural network classifier

IEEE Transactions on Information Technology in Biomedicine

Texture Analysis Using Modified Computational Model of Grating Cells in Content-Based Medical Image Retrieval

Medical Imaging and Informatics

Quantified Score

Hi-index	0.00

Visualization

Abstract

The objective of this work is developing a computer-aided diagnostic (CAD) system for focal liver lesions in CT images. The texture analysis methods are used for the classification of hepatocellular cancer and liver cysts. CT contrast enhanced images of 20 adult subjects with hepatocellular carcinoma or with non-parasitic solitary liver cyst were used as entry data. A total number of 130 spatial and second-order probabilistic texture features were computed from the images. Ensemble of Bayes classifiers was used for the tissue classification. Classification success rate was as high as 100% when estimated by leave-one-out method. This high success rate was achieved with as few as one optimal descriptive feature representing the average deviation of horizontal curvature computed from original pixel gray levels. This promising result allows further amplification of this approach in distinguishing more types of liver diseases from CT images.