论文部分内容阅读
目的探讨头颈部肿瘤分子生物纹理分析与生物靶区自适应的勾画方法。方法从肿瘤PET图像的共生矩阵中提取肿瘤分子生物方差纹理特征,然后结合肿瘤生物方差纹理特征,对之前的两个阶段自适应三维体生长方法进行改进,对头颈部肿瘤自适应生物靶区进行自适应勾画。结果联合鼻咽癌VAR纹理特征、PET SUV进行两级区域生长计算,一级区域生长阈值0.65,分割结果显示,分割轮廓性较强,所有区域均联通,经临床专家视觉评估,认定该分割结果合理、正确。结论改进后的生物靶区自适应勾画方法,可有效提高头颈部肿瘤生物靶区勾画的精确度。
Objective To investigate the molecular biological texture analysis of head and neck tumor and the outline of biological target zone adaptive method. Methods The bio-variance texture features of tumor were extracted from the co-occurrence matrix of tumor PET images. Based on the bio-variance texture features of tumor, the two-stage adaptive three-dimensional volume growth method was improved, and the adaptive target region of head and neck tumors Adaptive sketch. Results Combined with the VAR texture characteristics of nasopharyngeal carcinoma, the PET SUV was calculated for two-stage region growth. The first-level region growth threshold was 0.65. The segmentation results showed that the segmentation contour was strong and all the regions were unicomned. Reasonable and correct. Conclusion The improved method of self-adaptive mapping of biological target can effectively improve the accuracy of biological target mapping of head and neck tumor.