Fig. 1: MRI preprocessing approaches for 2D and 3D analysis. | Communications Medicine

Fig. 1: MRI preprocessing approaches for 2D and 3D analysis.

From: Volumetric spline-based Kolmogorov-Arnold architectures surpass CNNs, vision transformers, and graph networks for Parkinson’s disease detection

Fig. 1

a Example of single slices centered on the midbrain used individually to classify between PD and HC in the two-dimensional analyses. b Entire volumetric MRI scan used in the three-dimensional analyses. Gaps are inserted at regular intervals to help visualize the inner structure, but are not implemented in the analyses.

Back to article page