This paper presents a texture analysis method on digital chest radiograph to distinguish pneumoconiosis chest from normal chest. First, two lung fields are segmented from a digital chest X-ray image by the active shape model (ASM) method. Second, the chest image is preprocessed by multi-scale difference filter bank to enhance some detailed features of pneumoconiosis. Then the histogram features are extracted from each lung field, including mean, standard deviation, skew, kurtosis, energy and entropy. A support vector machine (SVM) classifier is utilized here to extract the discriminatory information through leave-one-out cross validation. Two experiments are conducted to evaluate the scheme by randomly selecting images from our chest database. The first test set includes 51 normal cases and 51 early stage cases; its classification result is sensitivity 91.1%, specificity 92.1%, and accuracy 91.6%;. The second test set includes 47 normal cases and 47 advanced stage cases; its classification result is sensitivity 93.6%, specificity 94.6%, and accuracy 94.1%. The analysis result shows that normal chest could be differentiated from pneumoconiosis chest distinctively.
[1]
Timothy F. Cootes,et al.
Active Shape Models-Their Training and Application
,
1995,
Comput. Vis. Image Underst..
[2]
Kunio Doi,et al.
Automatic detection of abnormalities in chest radiographs using local texture analysis
,
2002,
IEEE Transactions on Medical Imaging.
[3]
Kunio Doi,et al.
Computer-aided diagnosis in chest radiography
,
2007,
Comput. Medical Imaging Graph..
[4]
Vladimir N. Vapnik,et al.
The Nature of Statistical Learning Theory
,
2000,
Statistics for Engineering and Information Science.
[5]
Klinikum Westfalen,et al.
OCCUPATIONAL SAFETY AND HEALTH SERIES No . 22 ( Rev . ) GUIDELINES FOR THE USE OF ILO INTERNATIONAL CLASSIFICATION OF RADIOGRAPHS OF PNEUMOCONIOSES
,
2009
.
[6]
William B. Thompson,et al.
Computer Diagnosis of Pneumoconiosis
,
1974,
IEEE Trans. Syst. Man Cybern..