Regional analysis of cerebral hemodynamic changes during the head-up tilt test in Parkinson’s disease patients with orthostatic intolerance

Fig. 1. Approximate placement of the sources and detectors of the DOT system within the prefrontal area of the subject, along with the AAL rendered in AtlasViewer.

Fig. 2. Examples of hemodynamic changes (HbO-red, Hb-blue, and HbT-green) in the left and right superior, middle, and medial gyri during the HUTT in a patient with Parkinson’s disease diagnosed with orthostatic hypotension. The baseline period from which the relative hemodynamics were calculated is indicated on the graph. The three HUTT phases are additionally denoted on the graph: dynamic tilt (supine to 70°), static tilt (remain tilted at 70°), and post-tilt (70° to supine).

Fig. 3. Oxyhemoglobin time-series changes averaged over the (a) superior; (b) middle; and (c) medial frontal gyri for HCs (maroon) and patients with both Parkinson’s disease and orthostatic hypotension (pink). The dynamic tilt phase is highlighted with a gray box. The gyri are separated between the left (upper row) and right (bottom row) hemispheres. Significant differences (p<0.05) between the two groups are denoted with a green bar over the time points.

Fig. 4. Oxyhemoglobin time-series changes averaged over the (a) superior; (b) middle; and (c) medial frontal gyri in the PD-POS (red) and PD-NEG (blue) sub-groups. The dynamic tilt phase is highlighted with a gray box. The gyri are separated between the left (upper row) and right (bottom row) hemispheres. Significant differences (p<0.05) between the two groups are denoted with a green bar over the time points.

Fig. 5. Significant differences (p<0.05; marked in orange) between the left and right superior (first column), middle (second column), and medial frontal gyri (third column) for (a) HCs; (b) PD-OH patients; (c) PD-POS patients; and (d) PD-NEG patients.

Fig. 6. Example of a decision tree used to classify (a) PD-NOR and PD-OH patients and (b) PD-NEG and PD-POS patients. The feature used as a decision node to split the tree is represented with a green diamond. The dashed-line diamond represents a clinical feature and the solid-line diamond represents an oxygenation feature. Each path ends in a leaf node representing a class of patients with Parkinson’s disease.

Fig. 7. Example of a decision tree used to classify PD-NEG, PD-POS, and PD-OH patients. The feature used as a decision node to split the tree is represented by a green diamond. The dashed-line diamond represents a clinical feature and the solid-line diamond represents an oxygenation feature. Each path ends in a leaf node representing a class of patients with Parkinson’s disease.

Fig. 8. Accuracy of decision tree classification for three different trials: PD-NOR versus PD-OH; PD-NEG versus PD-POS; PD-OH versus PD-NEG versus PD-POS.

Fig. 9. Feature importance according to (a) oxygenation features and (b) clinical features of the three classification trials: PD-NOR versus PD-OH (gray bar); PD-NEG versus PD-POS (pink bar); and PD-NEG versus PD-POS versus PD-OH (yellow bar).

Fig. 10. Average feature importance of oxygenation features of the left and right gyri for (a) PD-NOR versus PD-OH; (b) PD-NEG versus PD-POS; and (c) PD-NEG versus PD-POS versus PD-OH.

Fig. 11. Box plot of the rate of HbO change for HCs, PD-POS, PD-NEG- PD-OH, and PD-NOR groups.