Skip to main content
. Author manuscript; available in PMC: 2018 Jan 1.
Published in final edited form as: Mov Disord. 2016 Oct 6;32(1):70–79. doi: 10.1002/mds.26806

FIG. 4. A linear support vector machine classifier provides high performance discrimination between the STN and SNr populations.

FIG. 4

A support vector machine (SVM) classifier was trained and tested on 155 randomly selected samples from the STN and all 155 samples from the SNr, using NRMS and the “100–150 Hz/5–25 Hz Power Ratio” features. The linear-kernel decision boundary is used to classify the trained data for the SNr (hollow square; green) and the STN (hollow triangle; blue); then new data points are classified as SNr (solid square; green) or STN (solid triangle; blue). Yellow circles (within the hollow squares and triangles) represent the support vectors defining the decision boundary between the STN and SNr samples.