Utility of 3D-T1WI and T2 FLAIR for diagnosis of Parkinson's disease using radionics
Objective Research on radionics with 3D-T1WI and T2-weighted fluid attenuated inversion recovery (T2 FLAIR) MRI images to investigate the changes of substantia nigra (SN) and red nucleus (RN) in patients with Parkinson's disease (PD) and differentiate PD from healthy controls. Method In this retrospective study,100 PD patients and 100 sex-and age-matched normal control (NC) subjects were enrolled from January 2013 to December 2019. All subjects underwent brain 3D-T1WI and T2 FLAIR imaging examination in 3T magnetic resonance imaging scanner. The regions of interest of the SN and RN were delineated by experienced radiologists. Three models including individual 3D-T1WI,individual T2 FLAIR and combination model were constructed by logistic regression analysis,and the receiver operating characteristic curve was used to evaluate the diagnostic efficiency. Spearman correlation tests were used to estimate the correlation between rad-scores and clinical scales in PD group. Result 1316 radiomic features based on 3D-T1WI and T2 FLAIR of left and right RN and SN were generated. Three radiomic models were performed good with the area under the receiver operator characteristic (ROC) curve (AUC) ranged from 0.82-0.94 in training set and 0.76-0.92 in testing set,and the left RN of combination model had best performance with AUC was 0.92 in testing set. The rad-score of right SN in T2 FLAIR model and left RN in 3D-T1WI model in PD group were negatively correlated with the Hamilton depression scale,and the rad-score of left RN in 3D-T1WI model was negatively correlated with the Hamilton anxiety scale (P<0.05). Conclusion The findings illustrated the radiomic features based on 3D-T1WI and T2 FLAIR of bilateral RN and SN had good performance to distinguish PD and healthy controls,and the combination model performed better than individual model. 3D-T1WI and T2 FLAIR have good potential for diagnosis of PD using radiomics.
万方数据
维普
