Oxygenation Encephalogram: Brain OS-MR Mirrors Cardiac OCG During Vasoactive Breathing

Background: Oxygenation-sensitive (OS) CMR has advanced myocardial oxygenation mapping, yet integrated, quantitative biomarkers of vascular function that translate between heart and brain remain scarce. In the heart, the oxygenation signal intensity (SI) across time, defined as oxygenation cardiogram (OCG), exhibits a characteristic hyperventilation (HV)-induced SI decrease (vasoconstriction) followed by a breath-hold (BH) SI increase (vasodilation), typically interpreted as a physiological oxygenation response (1). A critique, however, is that the OCG could partly reflect trans-planar motion (2). To address this and enable like-for-like comparison, we introduce the oxygenation encephalogram (OEG), a time-resolved OS-MR readout of brain gray matter, and test whether it displays similar trajectory under a standardized vasoactive breathing maneuver.

Methods: Thirty-nine participants (53.8 ± 9.2 y; 41% female) performed HV followed by a prolonged BH during OS-MR acquisition for the heart and brain. In-house MATLAB (MathWorks, Massachusetts, USA) pipelines sampled SI across time to generate OEG and OCG for Slice 1, Slice 2, and Total (per-subject mean of slices). The code also included baseline and motion corrections to reduce possible noise. From the OEG, HV cerebral oxygenation reserve (HV-CORE: % change between baseline and HV means) and Breathing induced CORE (B-CORE: % change between baseline and BH means) were extracted. From the OCG, breathing induced myocardial oxygenation reserve (B-MORE: % change between mean of last and 1^st cardiac cycles of BH) was extracted. Group values were summarized as median [IQR]. Directionality was tested with a two-sided sign test, and deviation from zero with the Wilcoxon signed-rank test. Brain–heart coupling was quantified using Spearman rank correlation (ρ).

Results: Visually, the OEG curves reproduced the decrease-increase pattern in brain (Figure 1). The HV decrease was quantified by a negative HV-CORE in Slice 1 −321% [−454, −74](sign-test p< 0.001; Wilcoxon p< 0.001), Slice 2 −354% [−560, −109] (p< 0.001; Wilcoxon p< 0.001), and Total: −387% [−541, −174] (p< 0.001; Wilcoxon p< 0.001). The BH increase was quantified by a positive B-CORE in Slice 1 +809% [+573, +1,072], (p< 0.001; Wilcoxon p< 0.001), Slice 2 +615% [0, +1,161], (p=0.001; Wilcoxon p=0.022), Total +751% [+305, +1,057] (p< 0.001; Wilcoxon p< 0.001). These magnitudes mirror our prior OEG work (3) and warrants further validation. B-MORE was +9.1 [−2.6, 19.1] (sign-test p=0.020; Wilcoxon p=0.017). B-CORE (slice 1) showed a moderate correlation with B-MORE (ρ=0.41, p=0.020) (Figure 2).

Conclusion: OEG extends the OCG paradigm to the brain and provides a unified, time-resolved measure of vasoactive reactivity, quantifying when and by how much SI changes. Demonstrating this decrease (HV) increase (BH) pattern in the brain, mirroring cardiac OCG, and quantifying its magnitude provides a temporal biomarker for cardio-cerebral vascular function while helping disentangle physiology from motion.

Figure 1. Brain OEG. Median (line) ± IQR (shaded) Δ% Signal Intensity across the breathing maneuver. Curves from all subjects are aligned at t=0 (vertical dashed line); Hyperventilation (left) shows a decrease in SI and Breath-hold (right) shows an increase.
Figure 2. Brain–heart coupling. Relationship between cardiac B-MORE and brain B-CORE (Slice 1). The line shows a least-squares fit with 95% CI. Spearman ρ=0.41, p=0.020. Positive slope indicates that greater cardiac vasodilation is associated with greater cerebral vasodilation.