Assessment of Heterogeneity Difference Between Edge and Core by Using Texture Analysis: Differentiation of Malignant From Inflammatory Pulmonary Nodules and Masses

详细信息    查看全文

• 作者：Shiteng Suo ; MS ; Jiejun Cheng ; MD ; Mengqiu Cao ; MD ; Qing Lu ; MD ; Yan Yin ; MD ; Jianrong Xu ; MD ; PhD ; ^{xujianrong_renji@163.com" class="auth_mail" title="E-mail the corresponding author} ; Huawei Wu ; MD ; ^{huaweiwu26@163.com" class="auth_mail" title="E-mail the corresponding author}
• 关键词：Heterogeneity ; texture analysis ; lung cancer ; inflammation ; contrast-enhanced CT
• 刊名：Academic Radiology
• 出版年：2016
• 出版时间：September 2016
• 年：2016
• 卷：23
• 期：9
• 页码：1115-1122
• 全文大小：1129 K

文摘

This study aimed to test the hypothesis that the heterogeneity difference between edge and core of lesions by using intensity and entropy features obtained from whole-lesion texture analysis on contrast-enhanced computed tomography (CT) may be useful for differentiation of malignant from inflammatory pulmonary nodules and masses.

Materials and Methods

In all, 48 single pulmonary nodules and masses were retrospectively evaluated. All lesions were histologically or clinically confirmed (malignancy: inflammation = 24:20). We utilized a newly introduced texture analysis method based on contrast-enhanced CT (first described by Grove et al.) that automatically divided the whole lesion volume into two regions: edge and core. Mean attenuation value (AV) and entropy of each region and also the whole lesion were evaluated separately. Each texture metric (absolute value for each region, and difference value defined as difference between edge and core) of malignant and inflammatory lesions were compared using Mann-Whitney U test. Individual image parameters were combined by using linear discriminant analysis. Receiver operating characteristic curves were generated to assess each texture metric and their combination for discriminating between the two entities.

Results

Mean AV difference and entropy difference were significantly higher in malignant lesions than in inflammatory lesions (4.71 HU ± 5.06 vs −1.53 HU ± 5.05, P < .001; 0.45 ± 0.23 vs 0.18 ± 0.30, P = .001). Receiver operating characteristic curves for individual mean AV difference and entropy difference provided relatively high values for the area under the curve (0.836 and 0.795, respectively). The combination of mean AV difference, entropy difference, and lesion volume improved the area under the curve to 0.864.

Conclusion

Heterogeneity difference between edge and core by using whole-lesion texture analysis on contrast-enhanced CT may be a promising tool for estimating the probability of malignancy in pulmonary nodules and masses.