基于组织相似度图谱磁共振灌注成像勾画脑胶质瘤边界

The cerebral glioma is one of the common malignant tumors that severely harm to human health. The margin of gliomas is hard to determine because of the infiltrative growth pattern. An important reason of postoperative recurrence is ascribe to the incomplete removal of the tumor. Furthermore, the brain function will stay in permanent loss or serious damage if the extent of surgical resection or the target area of postoperative radiotherapy is blindly expanded. Therefore, it is significantly important to accurately outline the boundary of pre-operative tumor with the help of medical imaging technique. In our early experiment, the project applicant has proposed a new method on the data analysis and image processing of magnetic resonance imaging (MRI), called the tissue similarity map (TSM). It is created by selecting a reference region from a given tissue of interest and comparing the mean squared errors (MSEs) gained from each pixel signal, which is able to distinguish the signal differences between any two tissues. 60 patients with brain gliomas will be selected as objectives in this project for the examination of MR perfusion imaging. The biological information such as hemodynamic of the tumor and the peri-tumoral normal tissue will be acquired and collected, which will be followingly processed by the data analysis and image processing via TSM method in order to outline the margin of tumor body. The accuracy of the TSM method in outlining the margin of glioma will be assessed in comparison with the conventional MR perfusion analysis and the pathological results from the MRI-guided surgery. The latter is regarded as the gold standard for the assessment in the study. TSM proposed in the study is expected to provide a new method for clinicians in accurately making the proper strategies of surgery and radiotherapy for the cerebral gliomas.

脑胶质瘤为危害人类健康的常见恶性肿瘤，浸润性生长方式导致其境界难辨，瘤灶切除不彻底是术后复发的重要原因，而盲目扩大手术切除范围及术后放疗靶区将会给患者造成永久性功能丧失或严重损害。因此，借助医学影像技术术前精确勾画肿瘤的边界意义重大。项目申请人研发了一种磁共振成像（MRI）数据分析和图像处理的新方法，称为组织相似度图谱（TSM），该方法通过选择特定组织的相关区域，比较所得数据的每个像素的信号之均方误差（MSE），可辨别任何两个组织之间的信号差异。本项目以60例脑胶质瘤患者为研究对象开展MRI灌注成像研究，采集肿瘤组织及瘤周正常脑组织的血流动力学等生物学信息，采用TSM方法进行数据分析和图像处理，勾画出肿瘤的边界。并以常规MRI灌注分析方法作对照，以MRI导航手术下取材行病理学检查之结果为金标准，评估TSM法勾画脑胶质瘤边界的准确性。该研究有望为临床精确制定脑胶质瘤手术及放疗方案提供新方法。

脑胶质瘤具有浸润性生长的特性，瘤灶难以彻底切除，是其术后复发的重要原因，因此术前精确勾画肿瘤边界具有重要的临床意义。本项目采用基于组织相似度图谱（TSM）对磁共振灌注成像数据进行分析，比较TSM算法与传统算法的异同，并用于勾画脑胶质瘤边界。分别于肿瘤边界内侧10mm及外侧10mm取材送检，以术中取材组织病理为标准，分析基于TSM算法MR灌注成像勾画脑胶质瘤边界的准确性。研究结果显示两种后处理算法都可以得到反应胶质瘤异常血流灌注的伪彩图，且两者在显示病灶形态及大小方面无明显差异，但TSM图上肿瘤组织与周围水肿带或正常脑组织之间对比更加明显，且TSM图的SNR较高。可见，应用rCBVTSM来评价脑胶质瘤的血流动力学特征是可行的。另外，将rCBVTSM与胶质瘤病理级别进行比较后发现可将其用于对胶质瘤进行分级诊断，肿瘤级别越高，rCBVTSM值越大。此外，rCBVTSM同样可用于鉴别高级别胶质瘤和单发脑转移瘤，高级别胶质瘤瘤周水肿区的rCBVTSM值明显高于单发脑转移瘤。共完成52例脑胶质瘤患者的术前MR灌注成像及术中导航穿刺取材，其中肿瘤内侧10mm证实为肿瘤组织为49例（94.2%），坏死组织3例（5.8%）；肿瘤外侧10mm证实存在肿瘤浸润9例（17.3%），未见肿瘤组织为43例（82.7%）。由此可见，基于TSM算法的MR灌注成像可较准确的反映出肿瘤浸润的边界，为临床制定精准的治疗（手术或放疗）提供参考。本项目对提高脑胶质瘤临床治疗疗效及安全性具有较高的理论意义和实用价值。