Journal: Neural Regeneration Research
Article Title: A multiple-tissue-specific magnetic resonance imaging model for diagnosing Parkinson’s disease: a brain radiomics study
doi: 10.4103/1673-5374.339493
Figure Lengend Snippet: Flowchart of the radiomics analysis framework. (A) The QSM, R2*, and T1-weighted images employed, and the features extracted from them. (B) The general linear model constructed from 121 normal controls to control for the influences of age and sex on the extracted original brain features. (C) The final informative radiomics features truncated through a data-driven feature selection. (D) The random forest framework used in the machine-learning training-testing cycles, which was parallelly tested on the patients with PD with different clinical statuses. Of note, independent external validation was conducted using an untouched database (database-106). EPD: Early PD; GLCM: Gray-Level Co-Occurrence Matrix; M-LPD: Moderate-to-late PD; NC: Normal controls; PD: Parkinson’s disease; PD-nonTD: non-tremor-dominant PD; PD-TD: Tremor-dominant PD; QSM: Quantitative susceptibility mapping.
Article Snippet: Second, three dimensional texture features were measured using the Gray-Level Co-Occurrence Matrix (GLCM) algorithm (Haralick et al., 1973) written in Matlab 2018a ( https://ww2.mathworks.cn/products/matlab.html ).
Techniques: Construct, Control, Selection, Biomarker Discovery