Scale-Space Representation of Lung HRCT Images for Diffuse Lung Disease Classification

摘要

A scale-space representation based on the Gaussian kernel filter and Gaussian derivatives filter is employed to describe HRCT lung image textures for classifying four diffuse lung disease patterns: normal, emphysema, ground glass opacity (GGO) and honey-combing. The mean, standard deviation, skew and kurtosis along with the Haralick measures of the filtered ROIs are computed as texture features. Support vector machines (SVMs) are used to evaluate the performance of the feature extraction scheme. The method is tested on a collection of 89 slices from 38 patients, each slice of size 512×512, 16 bits/pixel in DICOM format. The dataset contains 70,000 ROIs from slices already marked by experienced radiologists. We employ this technique at different scales and different directions for diffuse lung disease classification. The technique presented here has best overall sensitivity of 84.6% and specificity of 92.3%.