Optimized cerebral blood flow measurement in speckle contrast optical spectroscopy via refinement of noise calibration

TL;DR

This paper introduces an adaptive noise calibration refinement method for speckle contrast optical spectroscopy, significantly improving cerebral blood flow measurement accuracy, especially at low signal levels and larger source-detector distances.

Contribution

It presents an optimization-based framework that adaptively refines noise calibration to reduce artifacts and improve CBF measurement fidelity in SCOS.

Findings

Lowered the reliable CBF signal threshold from 97 to 26 electrons per pixel.

Validated on 10 human subjects showing improved measurement robustness.

Enhanced accuracy of CBF measurements at larger source-detector distances.

Abstract

Speckle contrast optical spectroscopy (SCOS) offers a non-invasive and cost-effective method for monitoring cerebral blood flow (CBF). However, extracting accurate CBF from SCOS necessitates precise noise pre-calibration. Errors from this can degrade CBF measurement fidelity, particularly when the overall signal level is low. Such errors primarily stem from residual speckle contrast associated with camera and shot noise, whose fluctuations exhibit a temporal structure that mimics cerebral blood volume (CBV) waveforms. We propose an optimization-based framework that performs an adaptive refinement of noise calibration, mitigating the CBV-mimicking artifacts by reducing the CBF-CBV waveform correlation. Validated on 10 human subjects, our approach effectively lowered the signal threshold for reliable CBF signal from 97 to 26 electrons per pixel for a 1920x1200 pixels SCOS system. This improvement enables more accurate and robust CBF measurements in SCOS, especially at large source-detector (SD) distances for deeper tissue interrogation.