论文部分内容阅读
目的在个体水平上应用脑磁敏感加权成像(SWI)的模式识别诊断帕金森病(PD)。方法分析36例可疑PD的震颤麻痹病人的脑SWI数据。这些病人都有①3TMRSWI扫描数据;②大脑123I-SPECT数据;③神经学检查数据,包括随访记录[16例PD,(67.4±6.2)岁,女11例;20例非典型震颤麻痹综合征病人,(65.2±12.5)岁,女6例]。对比两组之间的SWI值,并在个体水平上进行支持向量机(SVM)分析。结果简单目视分析,两组之间的SWI值无差异。但是组间分析显示,比较其他震颤麻痹征病人,PD病人双侧丘脑和左侧黑质的SWI值升高。逆向比较亦未超出阈值。在个体水平,SVM正确归类PD病人的正确率达到86%以上。结论在个体水平,尽管目视无法发现SWI值的变化,但SWI数据的SVM模式识别可以在各种震颤麻痹征者中准确地识别出PD病人。本项初步实验的结果为将来使用不同MR设备和MR参数进行大样本PD病人研究提供了保证。要点①MRI数据为研究PD提供了新视角。②目视SWI分析无法区分特发性PD与非典型PD。③SVM分析提高了检出特发性PD的准确性。④SVM分析为临床上诊断个体PD提供信息。⑤常规MRI可较容易地获得这些数据。
Objective To diagnose Parkinson’s disease (PD) using pattern recognition of brain magnetic susceptibility weighted imaging (SWI) at the individual level. Methods The brain SWI data of 36 suspicious PD patients with Parkinsonism were analyzed. These patients had ① 3TMRSWI scan data; ② 123I-SPECT brain data; ③ neurological examination data, including follow-up records [16 cases of PD, (67.4 ± 6.2) years old and 11 females; 20 patients with atypical tremor syndrome, (65.2 ± 12.5) years old, female 6 cases]. SWI values were compared between the two groups and support vector machine (SVM) analyzes were performed at the individual level. The results of a simple visual analysis, there was no difference between the two groups of SWI values. However, the intergroup analysis showed an increase in SWI values in bilateral thalamus and left substantia nigra in PD patients compared with those in other tremor paralysis patients. The reverse comparison did not exceed the threshold. At the individual level, the correct rate of SVM correctly classified PD patients reached more than 86%. Conclusions At the individual level, despite the inability to visually detect changes in SWI values, SVM pattern recognition of SWI data can accurately identify PD patients among a variety of paralyzed paralyze types. The results of this preliminary experiment provide a guarantee for the future study of large sample PD patients using different MR devices and MR parameters. Points ①MRI data provide a new perspective for the study of PD. ② visual SWI analysis can not distinguish between idiopathic PD and atypical PD. ③ SVM analysis to improve the detection of idiopathic PD accuracy. ④ SVM analysis for the clinical diagnosis of PD to provide information. ⑤ conventional MRI can easily obtain these data.