Introduction

Non-invasive measurements of perfusion metrics, such as cerebral blood volume (CBV) and cerebral blood flow (CBF) are essential for effectively monitoring disease progression and treatment response. To achieve these non-invasive measurements, we adopted a BOLD dynamic susceptibility contrast (DSC) method with transient hypoxia³. In BOLD-DSC MRI, it is crucial to maximize hypoxia-induced signal changes and accurately determine arterial input functions. Therefore, optimizing the appropriate echo time (TE) becomes vital, as it impacts both the baseline signal-to-noise ratio and the magnitude of signal changes. A shorter TE results in reduced hypoxia-induced changes, while a longer TE reduces the baseline SNR and increases the risk of arterial signal saturation^1,4. To address this, we systematically varied TE within the range of 11.57 ms to 20 ms across six mice, allowing us to measure hypoxia-induced signal changes in arterial, venous, and somatosensory tissue voxels, and accurately quantify perfusion metrics.

Methods

Hypoxic Gas Stimulation
Transient anoxic stimulus was delivered using a block design paradigm of 60s rest (40% O₂/ 60% N₂) and 5s stimulation (100% N₂) alternatively repeated five times (Figure 1.A).
BOLD acquisitions
BOLD MRI studies were acquired on a 9.4T system using GE-EPI sequence with varying echo times TR/TEs =1000/11.5, 13, 15, 17 and 20ms, FA=50°, 156x156x500 μm³ , 20 slices. Hypoxic stimulus was administered under 1.5% Isoflurane in 6 animals.
Data analysis
An automatic algorithm was used for selecting candidate arterial and venous voxel, while tissue voxel was chosen from the somatosensory region. The noise level was given by the standard deviation (SD) of the baseline signal. Baseline SNR was calculated SNR= S₀/SD (S₀: mean baseline signal intensity). Hypoxia-induced ΔR^*₂ at different TEs were calculated based on: ∆S/S₀=-TE × ΔR^*₂ (∆S: hypoxic-induced signal change). To quantify perfusion values from dynamic hypoxia-induced BOLD responses, we adopted the DSC theory³.

Results

Longer TEs cause arterial and venous peak signal saturations
The baseline values were 24.07±2.17 ms for the artery, 9.23±0.59 ms for the vein, and 23.81±3.00 ms for the tissue (Fig. 1B). The effects of longer TEs on baseline signal and SNR were clearly observed in artery, vein, and tissue voxels with longer TEs reducing the baseline SNR (Figs. 1A, C). We observed peak saturations in the arterial voxel at TEs of 15, 17 and 20 ms (Fig. 1D-i), and in the venous voxel at TEs of 17 and 20 ms (Fig. 1D-ii).
Peak saturations lead to underestimations of arterial input function
The signal time curves of the artery, vein and tissue were then converted to concentration time curves (ΔR^*₂). Noisy signal time curves from the artery and vein, acquired at longer TEs of 15ms, 17ms and 20 ms resulted in lower concentration time curves (Fig. 2A-i, ii), indicating that peak saturations at longer TEs cause underestimation of the arterial and venous concentration time curves. In contrast, at shorter TE values of 11.5 ms and 13 ms, significantly higher and similar concentration time curves were induced in the artery and vein. No significant differences were observed in the ΔR^*₂ curves obtained at all different TEs in tissue (Fig. 2A-iii). To remove partial volume effect of arterial voxel, the arterial ΔR^*₂ response was then normalized by the venous ΔR^*₂ response, resulting in a corrected AIF (Fig. 2B-i). We observed that the resulting corrected AIFs will be underestimated when the arterial and venous peak saturations occurred at longer TEs (Fig. 2B, i-ii).
Overestimations of perfusion metrics due to AIF underestimations
Hypoxia-induced GE-BOLD MRI signal changes were measured under various TEs, and subsequently converted into absolute CBV and CBF values with animal-specific corrected AIF³. To enhance clarity regarding the general patterns at these five TEs, we presented three selected slices of group-averaged maps for better visualization (Figs. 3, D, F). Our CBF values measured at TEs of 11.5 ms and 13 ms were similar and consistent with previous publications conducted under the same anesthetics⁵. Peak saturation of the AIF is a major source of systematic errors in the quantification of CBV and CBF in MRI^2,4. Consequently, the calculated CBV and CBF under longer TEs (15 ms, 17 ms, and 20 ms) were increased (overestimated) as compared with those values of short TE (Figs. 3D-F).

Discussion and Conclusions

Our data demonstrates that a shorter TE results in fewer hypoxia-induced signal changes, while a longer TE leads to decreased baseline SNR and an increased risk of arterial signal saturation. The peak saturation of the AIF results in significant errors in the quantification of CBV and CBF and needs to be considered in DSC perfusion studies.

Acknowledgements

This research was supported by the Institute of Basic Science (IBS-R015-D1).