School of Biomedical Engineering, Capital Medical University, You An Men, Beijing, China.
PLoS One. 2012;7(6):e39936. doi: 10.1371/journal.pone.0039936. Epub 2012 Jun 28.
The purpose of this paper is to report the noninvasive imaging of hepatic tumors without contrast agents. Both normal tissues and tumor tissues can be detected, and tumor tissues in different stages can be classified quantitatively. We implanted BEL-7402 human hepatocellular carcinoma cells into the livers of nude mice and then imaged the livers using X-ray in-line phase-contrast imaging (ILPCI). The projection images' texture feature based on gray level co-occurrence matrix (GLCM) and dual-tree complex wavelet transforms (DTCWT) were extracted to discriminate normal tissues and tumor tissues. Different stages of hepatic tumors were classified using support vector machines (SVM). Images of livers from nude mice sacrificed 6 days after inoculation with cancer cells show diffuse distribution of the tumor tissue, but images of livers from nude mice sacrificed 9, 12, or 15 days after inoculation with cancer cells show necrotic lumps in the tumor tissue. The results of the principal component analysis (PCA) of the texture features based on GLCM of normal regions were positive, but those of tumor regions were negative. The results of PCA of the texture features based on DTCWT of normal regions were greater than those of tumor regions. The values of the texture features in low-frequency coefficient images increased monotonically with the growth of the tumors. Different stages of liver tumors can be classified using SVM, and the accuracy is 83.33%. Noninvasive and micron-scale imaging can be achieved by X-ray ILPCI. We can observe hepatic tumors and small vessels from the phase-contrast images. This new imaging approach for hepatic cancer is effective and has potential use in the early detection and classification of hepatic tumors.
本文旨在报告无对比剂的肝脏肿瘤无创成像。既能检测正常组织，又能定量分类不同阶段的肿瘤组织。我们将 BEL-7402 人肝癌细胞植入裸鼠肝脏，然后使用 X 射线线衬度相位对比成像（ILPCI）对肝脏进行成像。提取基于灰度共生矩阵（GLCM）和双树复小波变换（DTCWT）的投影图像纹理特征，以区分正常组织和肿瘤组织。使用支持向量机（SVM）对不同阶段的肝肿瘤进行分类。接种癌细胞 6 天后处死的裸鼠肝脏图像显示肿瘤组织弥漫分布，但接种癌细胞 9、12 或 15 天后处死的裸鼠肝脏图像显示肿瘤组织中有坏死肿块。基于 GLCM 的正常区域纹理特征的主成分分析（PCA）结果为正，而肿瘤区域的结果为负。基于 DTCWT 的正常区域纹理特征的 PCA 结果大于肿瘤区域。低频率系数图像中纹理特征的值随肿瘤的生长而单调增加。可以使用 SVM 对不同阶段的肝肿瘤进行分类，准确率为 83.33%。X 射线 ILPCI 可以实现非侵入性和亚微米成像。我们可以从相衬图像中观察到肝脏肿瘤和小血管。这种新的肝癌成像方法有效，具有在早期检测和分类肝肿瘤方面的应用潜力。
