Spectral fiber-photometry derives hemoglobin concentration changes for accurate measurement of fluorescent sensor activity

Weiting Zhang, Tzu-Hao Harry Chao, Yue Yang, Tzu-Wen Wang, Sung-Ho Lee, Esteban A. Oyarzabal, Jingheng Zhou, Randy Nonneman, Nicolas C. Pégard, Hongtu Zhu, Guohong Cui, Yen‐Yu Ian ShihbioRxiv 2021 · DOI

Abstract

Fiber-photometry is a powerful tool for recording neural activity via fluorescent sensors, yet it is frequently confounded by hemoglobin-absorption artifacts. Because hemoglobin (Hb) concentrations fluctuate dynamically with neural activity, these "hemodynamic ghosts" can lead to erroneous interpretations of sensor data. This study introduces a spectrally resolved fiber-photometry platform and a computational framework using the Generalized Method of Moments (GMM) to quantify and correct for Hb-absorption. By leveraging activity-independent fluorescent proteins, the method successfully isolates true neural signals and provides high-sensitivity measurements of oxy- and deoxy-hemoglobin concentration changes, revealing significant regional variations in neurovascular coupling.

Introduction: The "Blood Tax" in Photometry

The brain's dense vascularization presents a significant challenge for optical recordings. Hemoglobin, in both its oxygenated (HbO) and deoxygenated (HbR) forms, is a potent light absorber across the visible spectrum. During neural activity, the resulting hemodynamic response—changes in blood volume and oxygenation—modulates the amount of light reaching and returning from fluorescent sensors. This "blood tax" can manifest as pseudo-undershoots or even flip the polarity of signals like GCaMP, potentially leading researchers to misinterpret vascular changes as neural inhibition or activation.

Methods: Spectral Resolution and Pathlength Modeling

To disentangle these signals, the researchers transitioned from traditional single-band photometry to spectrally resolved recording. This approach captures the entire emission spectrum, providing the data density needed to solve for multiple variables simultaneously.

Experimental setup and mechanism of hemoglobin attenuation in fiber-photometry.

The methodology relies on three pillars:

  1. Monte Carlo Simulations: These models determine the average traveling pathlengths ($X$) for excitation and emission light within brain tissue, which is essential for calculating the magnitude of absorption.
  2. Static References: By co-expressing activity-independent proteins (e.g., EYFP), researchers can identify fluctuations caused solely by hemoglobin absorption.
  3. GMM Framework: A computational model solves for changes in HbO and HbR concentrations using dozens of spectral data points, allowing for precise signal restoration.

Results: Correcting the 'Hemodynamic Ghost'

The proposed GMM method demonstrated high accuracy in "flattening" the signals of static proteins during stimulation, effectively removing the absorption artifacts. For instance, a 20% increase in cerebral blood volume (CBV) was found to cause a ~4% decrease in green fluorescence, a massive error that the correction method successfully eliminated.

Validation of the spectral correction method using static EYFP and derived hemoglobin concentration changes.

Beyond correction, the spectral method proved more sensitive than MRI-based CBV measurements. It successfully detected the "initial dip" in deoxygenated hemoglobin—a transient rise in HbR preceding the surge of oxygenated blood—which is typically invisible in raw photometry data. Furthermore, the study revealed that the Hemodynamic Response Function (HRF) varies significantly between brain regions, such as the motor cortex (M1) and the striatum (CPu), suggesting that neurovascular coupling is not a uniform process across the brain.

Discussion and Future Directions

While spectral photometry offers a robust solution to hemoglobin artifacts, it currently requires more complex hardware (spectrometers) and interleaved excitation pulses that can impact temporal resolution. Future work aims to integrate motion correction into this pipeline, potentially through a "triple-color" approach using a far-red static reference. By providing a clearer view of both neural sensors and hemodynamics, this platform enhances the reliability of fiber-photometry in diverse experimental contexts.

Summary updated

Abstract

Summary Fiber-photometry is an emerging technique for recording fluorescent sensor activity in the brain. However, significant hemoglobin-absorption artifacts in fiber-photometry data may be misinterpreted as sensor activity changes. Because hemoglobin exists in nearly every location in the brain and its concentration varies over time, such artifacts could impede the accuracy of many photometry recording results. Here we present a novel use of spectral photometry technique and propose computational methods to quantify photon absorption effects by using activity-independent fluorescence signals, which can be used to derive oxy- and deoxy-hemoglobin concentration changes. Following time-locked neuronal activation in vivo , we observed that a 20% increase in CBV contributes to about a 4% decrease in green fluorescence signal and a 2% decrease in red fluorescence signal. While these hemoglobin concentration changes are often temporally delayed than the fast-responding fluorescence spikes, we found that erroneous interpretation may occur when examining pharmacology-induced sustained activity changes, and in some cases, hemoglobin-absorption could flip the GCaMP signal polarity. We provided hemoglobin-based correction methods to restore fluorescence signals across spectra and compare our results against the commonly used regression approach. We also demonstrated the utility of spectral fiber-photometry for delineating brain regional differences in hemodynamic response functions. Highlights Hemoglobin-absorption compromises fiber-photometry recording in vivo Spectral photometry allows quantification of hemoglobin concentration changes for correction The proposed platform allows measuring regional differences in neurovascular transfer function
BibTeX
@misc{zhang2021spectral,
  author = {Weiting Zhang and Tzu-Hao Harry Chao and Yue Yang and Tzu-Wen Wang and Sung-Ho Lee and Esteban A. Oyarzabal and Jingheng Zhou and Randy Nonneman and Nicolas C. Pégard and Hongtu Zhu and Guohong Cui and Yen‐Yu Ian Shih},
  title = {{Spectral fiber-photometry derives hemoglobin concentration changes for accurate measurement of fluorescent sensor activity}},
  year = {2021},
  doi = {10.1101/2021.08.23.457372},
  howpublished = {bioRxiv},
}
Figure