Optimal binning and ROI size of the automatic classification system for differentiation between obstructive lung diseases on the basis of texture features at HRCT

概要

論文の詳細を見る
To find optimal binning and ROI size of the automatic classification system for differentiation between obstructive lung diseases on the basis of textural analysis at HRCT. The images were selected from HRCT images obtained in 17 healthy subjects (NL, n=66), 26 patients with bronchiolitis obliterans (BO, n=69), 28 patients with mild centrilobular emphysema (CLE, n=64), and 21 patients with panlobular emphysema or severe centrilobular emphysema (PLE, n=62). Images were acquired with 0.75mm collimation and sharp kernel (B70f) (Sensation 16; Siemens). For each image, an experienced radiologist selected a rectangular ROIs with variable size (16x16, 32x32, 64x64 pixels), representing each disease or normal lung. Areas with HU between -400 and -1024 were segmented. Histogram features (mean, SD, skewness, kurtosis), gradient features (mean, SD), run length encoding (short and long primitive emphasis), and co-occurrence matrix (angular second moment, contrast, correlation, entropy, inverse difference momentum) were employed. To find optimal binning, variable binning size (Q: 16, 32, 64, 128, 144, 196, 256) were tested. For automated classification, a Bayesian classifier was implemented. To assess the performance and cross-validation of the system, 5 folding method was used. Overall sensitivities of the system with the combination of variable ROIs, binning sizes were calculated and compared. The optimal combination of ROI size, binning size for run length encoding, and binning size for co-occurrence matrix was 64x64 pixels, 196 and 32 separately. When optimal binning and other parameters were set, overall sensitivity of the classifier was 87.95%. The sensitivity and specificity of the system for each class were as follows: NL, 90%, 95.4%; BO, 78.3%, 95.7%, CLE, 83.3%, 92.1%; and PLE 94.9%, 98.7%, respectively. We determined the optimal binning and ROI size of the automatic classification system for differentiation between obstructive lung diseases on the basis of texture features at HRCT
社団法人電子情報通信学会の論文
2007-01-19

Optimal binning and ROI size of the automatic classification system for differentiation between obstructive lung diseases on the basis of texture features at HRCT

スポンサーリンク

概要

著者

関連論文

スポンサーリンク