Objective: This present study was aimed to investigate the potential value of the substantia nigra (SN) radiomic feature for the prediction of PD progression.

Background: Although the negative of swallow tail sign is promising for the diagnose of Parkinson’s disease (PD), it is still a great challenge to use MRI to detect the progress of this disease.

Method: We retrospectively included the PD patients and sex, aged matched controls between Jan 2016 and Nov 2019 in our Movement Disorder Center who underwent 3T MR imaging of the brain including SWI sequence. A total of 328 radiomics features were extracted from the SN based on the susceptibility weighted imaging (SWI) of 95 PD patients and 95 controls. The maximum correlation minimum redundancy algorithm (mRMR) and the least absolute shrinkage and selection operator (LASSO) were conducted to filter and choose the optimized subset of features. After that, the most predictive subset of feature was chosen and the correlation analyze of these predictive features with clinical factor was conducted. At last, we used ROC analysis to evaluate the performance of the model built by linear regression.

Results: In the training set of the model based on LASSO dimensionality construction, AUC = 0.778 and in the validation set, AUC = 0.606. Based on Youden Index, the sensitivity and specificity of radiomics in the validation set were 64% and 72.4% for PD vs controls. The radiomics features showed a moderate correlation with H-Y stage (Spearman corresponding coefficient =0.49).

Conclusion: Our study indicated that radiomics based on SWI could provide the potential value of for the prediction of PD progression.

« Back to MDS Virtual Congress 2020

MDS Abstracts - https://www.mdsabstracts.org/abstract/application-of-substantia-nigra-feature-extraction-in-the-progression-of-parkinsons-disease-based-on-magnitude-image-of-susceptibility-weighted-imaging/